UvA-DARE (Digital Academic Repository) The Application of ... · 1.3 A D utch perspective 3 1.4 T his study 4 2 A rch aeological p rosp ection in T h e N eth erlan d s 7 2.1 Introduction

UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl)

UvA-DARE (Digital Academic Repository)

The Application of Magnetic Methods for Dutch Archaeological Resource Management

Kattenberg, A.E.

Link to publication

Citation for published version (APA):Kattenberg, A. E. (2008). The Application of Magnetic Methods for Dutch Archaeological ResourceManagement. Amsterdam: Amsterdam Institute for Geo and Bioarchaeology, Vrije Universiteit.

General rightsIt is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s),other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulationsIf you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, statingyour reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Askthe Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam,The Netherlands. You will be contacted as soon as possible.

Download date: 23 Oct 2020

https://dare.uva.nl/personal/pure/en/publications/the-application-of-magnetic-methods-for-dutch-archaeological-resource-management(073edc09-5549-4b40-8396-afa7e03504bc).html


Alette E. Kattenberg

Geoarchaeological and Bioarchaeological Studies Volume 9, 2008

ii

Geoarchaeological and Bioarchaeological Studies is published by the Institute for Geo and Bio-archaeology (Faculty of Earth and Life Sciences) of the Vrije Universiteit in Amsterdam, Netherlands. The series presents Ph.D. studies, scientific reports of contract research, conference proceedings, etc., in the field of archaeological science which have been done by, or were performed under the supervision of the Institute for Geo- and Bioarchaeology. Relevant studies carried out by other organisations appealing to a broadly interested readership in archaeological science are invited to submit their manuscripts to the series. Contributions should be written in English. Manuscripts in Dutch will be accepted only as an exception. Editorial Board H. Kars, A.M.J. de Kraker and S.J. Kluiving Pre-press Vrije Universiteit, Amsterdam Coverdesign Bert Brouwenstijn, Vrije Universiteit, Amsterdam Printed by PrintPartners Ipskamp B.V., Amsterdam Distribution You may order volumes of the series by sending an e-mail message to the secretariat of the Institute: [email protected]. Price of the present volume € 35.00 including postage and handling, but excl. VAT. This volume presents the original version of the Ph.D. thesis of Ms. A.E. Kattenberg and has been reviewed by the Doctorate Commission composed of G.J. Borger (Amsterdam), J. Fassbinder (München), R.M. van Heeringen (Amersfoort), J. Sevink (Amsterdam), and V.V. Stissi (Amsterdam). All rights reserved © Institute for Geo and Bioarchaeology, Vrije Universiteit, Amsterdam ISBN 978-90-77456-09-5 ISSN 1571-0750 Photo on cover: Magnetometer survey of the drowned village of Valkenise on the mudflats of the Westerschelde (photo by author). Data: Results of the magnetometer survey in Smokkelhoek.

iii


ACADEMISCH PROEFSCHRIFT

ter verkrijging van de graad van doctor

aan de Universiteit van Amsterdam

op gezag van de Rector Magnificus

prof. dr. D.C. van den Boom

ten overstaan van een door het college voor promoties ingestelde

commissie, in het openbaar te verdedigen in de Agnietenkapel

op vrijdag 6 juni 2008, te 12.00 uur

door Alette Else Kattenberg

geboren te Amsterdam

iv

Promotores: prof. dr. J.H.F. Bloemers

prof. dr. H. Kars Co-promotor: dr. A. Schmidt Faculteit der Geesteswetenschappen

v

Contents Preface and acknowledgements ix 1 Introduction 1 1.1 A magical method? 1 1.2 Aims and objectives 3 1.3 A Dutch perspective 3 1.4 This study 4 2 Archaeological prospection in The Netherlands 7 2.1 Introduction 7 2.2 The history of archaeological prospection in The Netherlands 9 2.3 The history of archaeological prospection from an international perspective 12 2.4 Archaeological prospection in The Netherlands 13 2.5 Conclusion 14 3 The principles of the application of magnetic methods for archaeological prospection focused on The Netherlands 15 3.1 Introduction 15 3.2 Different types of magnetism 15 3.3 Magnetic susceptibility 16 3.4 Soil iron oxides 17 3.5 Enhancement of magnetic susceptibility 18 3.6 Magnetic susceptibility in The Netherlands 20 3.7 Induced magnetization 22 3.8 Remanent magnetization 24 3.9 Magnetic anomalies 24 3.9.1 Direction of magnetization 24 3.9.2 Size and shape 25 3.10 Post depositional processes 27 3.11 Conclusion 28 4 Methodology 31 4.1 Choice of sites 31 4.2 Inventory of sites 32 4.3 Instruments 35 4.3.1 Magnetometer survey 35 4.3.2 Fluxgate magnetometer 35 4.3.2.1 Geoscan FM36 fluxgate gradiometer 35 4.3.2.2 Bartington GRAD601 fluxgate gradiometer 36 4.3.3 Magnetic susceptibility 36 4.3.3.1 Agico KLY-2 magnetic susceptibility bridge 36 4.3.3.2 Bartington MS2B susceptibility meter 37 4.3.3.3 Fractional conversion 37

vi

4.3.4 Thermomagnetic, IRM and ARM measurements 37 4.3.4.1 Thermomagnetic measurements 37 4.3.4.2 Anhysterectic remanent magnetization (ARM) measurements 38 4.3.4.3 Isothermal remanent magnetization (IRM) measurements 38 4.4 Software 38 4.4.1 Geoplot 38 4.4.2 Archeosurveyor 39 4.4.3 Further software 39 5 Estuarine deposits 41 5.1 Introduction 41 5.2 The mapping of a peat-extraction landscape 41 5.2.1 Introduction 41 5.2.2 The problem 41 5.2.3 Methods for the investigation of former peat extraction 43 5.2.4 The use of geophysical methods for the investigation of former peat extraction 44 5.2.5 Sites 45 5.2.6 Methodology 46 5.2.7 Results 47 5.2.8 Discussion 48 5.2.9 Conclusion 49 5.3 Archaeological prospection of the Dutch estuarine landscape by means of magnetic methods 50 5.3.1 Introduction 50 5.3.2 Archaeology and geology 51 5.3.3 Methodology 51 5.3.4 Results 52 5.3.5 Discussion 55 5.3.6 Conclusion 56 5.4 Iron compounds from archaeological features in estuarine deposits, examples from The Netherlands 57 5.4.1 Introduction 57 5.4.2 Background 57 5.4.2.1 Magnetic susceptibility 57 5.4.2.2 Enhancement of topsoil 59 5.4.3 Methods and materials 60 5.4.4 Results 67 5.4.5 Discussion 71 5.4.6 Conclusion 75 6 Wind blown and fluvial deposits 77 6.1 Wind blown sands 77 6.1.1 Magnetic susceptibility of the cover sands 78 6.1.2 Magnetic anomalies in the cover sands 80 6.1.3 Masking 80 6.1.4 Variability 82 6.1.5 Magnetic susceptibility of coastal dunes 84 6.1.6 Magnetic anomalies in coastal dunes 86 6.2 Loess 86 6.2.1 Magnetic susceptibility 87 6.2.2 Magnetic anomalies 87 6.3 Fluvial deposits 88 6.3.1 Magnetic susceptibility 89 6.3.2 Magnetic anomalies 90 6.4 Conclusions 91

vii

7 Magnetic mapping of archaeological features in The Netherlands 95 7.1 Settlements 95 7.1.1 Pits 95 7.1.2 Ditches 97 7.1.3 Walls 98 7.1.4 Wells 99 7.2 Off-site 99 7.2.1 Plough marks 99 7.2.2. Ditches 100 7.2.3 Watering pits 101 7.3 Funerary structures 101 7.3.1 Graves 101 7.3.2 Tumuli (ring gullies) 102 7.4 Industrial 102 7.4.1 Peat ties and extraction pits 102 7.4.2 Furnaces 103 7.5. Infrastructure 104 7.5.1 Roads 104 8 Discussion 105 8.1 Methodology 105 8.2 Contrasts in magnetic susceptibility 106 8.3 Estuarine and marine deposits 107 8.4 Wind blown deposits 108 8.5 Fluvial deposits 109 8.6 Magnetic anomalies unrelated to the geological environment 109 9 Conclusion 111 9.1 Principles of magnetic prospection 111 9.2 Magnetic contrasts in different geogenetic environments 112 9.3 Masking and variability 112 9.4 Iron sulphide formation 113 9.5 Returning to the objectives 114 9.6 The integration of magnetometry into ARM 114 9.7 The application of magnetic methods 115 Appendix I Fact sheets 117 Introduction 117 1 Beugen 117 2 Beugen 118 3 Broekpolder 123 4 Deil 127 5 Den Dolder 128 6 Breda 130 7 Borgharen 133 8 Harnaschpolder 136 9 Harnaschpolder, north, south and east 139 10 Heeten 143 11 Kolhorn 145 12 Limmen 147 13 Meerssen 151 14 Meteren 154 15 Oostende (Belgium) 157 16 Ossenisse 158

viii

17 Poeldijk 161 18 Polre 163 19 Raalte 165 20 Slabroek 166 21 Smokkelhoek 168 22 Spalding (United Kingdom) 171 23 Stede Broec 175 24 Steenbergen 177 25 Swalmen 179 26 Uitgeest 180 27 Uitgeesterbroek 181 28 Valkenisse 183 29 Wijk bij Duurstede 186 30 Zaltbommel 187 31 Zwaagdijk Oost 189 Appendix II Laboratory data 193 1 Heating experiments 193 2 Curie Balance measurements 196 2.1 Introduction 196 2.2 Sample selection 197 2.3 Methodology 197 2.4 Results 198 3 IRM component analysis 207 Appendix III Hand augering data 209

References 219 Samenvatting 225

ix

Preface and acknowledgements This thesis is the result of research into the application of magnetic methods for Dutch archaeological resource management and forms part of a larger study: the development of a method for the pros-pection, characterization and inspection of archaeological sites and landscapes in the Netherlands, using a multidisciplinary approach. The other two sections of this larger study concern a thesis on the development of chemical prospection methods by Stijn Oonk1 and a state of the art of remote sensing techniques by Chris Sueur2. This overall project has been carried out with financial support of the Netherlands Organisation for Scientific Research (NWO) in the programme Protecting and Deve-loping the Dutch Archaeological-Historical Landscape (Bodemarchief in Behoud en Ontwikkeling), the Amsterdam Archaeological Center of the University of Amsterdam, and the Rijksdienst voor Archeologie, Cultuurlandschap en Monument at Amersfoort. In the preamble to this program the urgent need for a larger toolbox for archaeological prospection in The Netherlands and for the integration of available methods was expressed. The studies that were to be conducted within the framework of this program needed to be tailored to be relevant for archaeo-logical resource management (ARM), rather than having only purely scientific goals. For this particular section of the project, many of the field studies that were carried out were inte-grated into existing archaeological projects, most of which were initiated to investigate the archaeo-logical record prior to soil disturbing activities. These close encounters with the practical side of ARM have helped to focus the results of this study to the target group, i.e. to parties that are looking to implement new technology in order to improve the quality of archaeological survey results. Readers with an archaeological background could read Chapter 1, as a general introduction, followed by Chapter 3, in which the principles behind magnetic methods are explained. Archaeological geophysicists would benefit from reading Chapter 1, the general introduction, and Chapter 2, which describes the position of geophysical methods in The Netherlands. Chapter 5 and 6 form the core of this study, whereas Chapter 7 provides an overview of magnetic anomaly shapes that were recorded in the magnetometer surveys, which could be used as reference material for the interpretation of future magnetometer survey results. First and foremost, I would like to thank Prof. Henk Kars for giving me the opportunity and the resources to conduct my research at the Institute for Geo- and Bioarchaeology at the VU University Amsterdam. As my primary supervisor, and through his trust he has coached me to develop my teaching and project management skills. I am much indebted to Prof. Tom Bloemers (Amsterdam Archaeological Center, University of Amsterdam), who has supported me not only during my PhD research, but throughout my archaeo-logical career, which started in 1993. His enthusiasm and perseverance have always been an example for me. I wish to express my heartfelt thanks to Dr. Armin Schmidt (Department of Archaeological Sciences of the University of Bradford) for the enjoyable and interesting discussions that we have had during the years. I feel that they have kept me on the right track. I would like to extend my sincere gratitude to the people at the Paleomagnetic Laboratory Fort Hoofddijk of the Universiteit van Utrecht, especially to Dr. Mark Dekkers and Mr. Tom Mullender, for giving me the opportunity to work in the Walhalla of magnetic research and for their help and discussions, which gave a new dimension to this PhD research. 1 Institute for Geo- and Bioarchaeology, Vrije Universiteit Amsterdam. 2 RAAP Archaeological Consultancy; present position at Vestigia, Archeologie en Cultuurhistorie, Amersfoort.

x

In the University of Bradford I would like to thank Gerry McDonnell and Stuart Fox for supervising and assisting in the heating experiment which I carried out in their laboratory facilities. At the Faculty of Earth Sciences of the VU University Amsterdam I would like to thank Martin Konert for giving me access to the sediment laboratory. Dr. Chris Gaffney I would like to thank for the thought-provoking discussions that we have had and his sincere interest in this study, and the other people at GSB Prospection for their support and for the window into day-to-day magnetometry. The magnetometer which I used in the first year of this study was borrowed from RAAP Archaeo-logical Consultancy, for which I would like to thank the people that made this possible. I would like to thank David Wilbourn (DW Consulting) for allowing me to use his software Archeosurveyor, for the 24/7 technical support that he provided and for giving me a hand in the field. There are many people that have facilitated the project by granting or organizing field access, by allowing me to take samples in excavation trenches, by taking samples for me, by laying out measure-ment systems, by field discussions and by sharing their field data; Ria Berkvens, Gerard Boreel, Epko Bult, Gilbert Busé, Menno Dijkstra, Erik Drenth, Jeroen Flamman, Bas van Geel, Tiziano Goossens, Klaas Greving, Tessa de Groot, Robert van Heeringen, Mieke Hissel, Bram Jansen, Mariëlle Kene-mans, Sjoerd Kluiving, Dicky de Koning, Jan de Koning, Adrie de Kraker, Fedor van Kregten, Tom Lane and the other APS people, Silke Lange, Heleen van Londen, Axel Müller, the people of the Planarch Project, Brigitte Quadflieg, Elma Schrijer, Maaike Sier, Carla Soonius, Eveline van de Steen, Liesbeth Theunissen, Adrie Ufkes, Henk van der Velde and Erik Verhelst, and all the land owners, a great thanks for making this project possible. Others have helped me while collecting data in the field: Gerard Aalbersberg, Geuch de Boer, Mieke Hissel, Bram Jansen, Karel-Jan Kerckhaert, Sjoerd Kluiving, Martijn Konst, David Wilbourn, Ian Wilkins and the students from the archaeological prospection methods course in 2004, 2005 and 2006 at VU University. The other participants in the NWO-BBO projects and Mies Wijnen (NWO) are kindly acknowledged. And finally I would like to thank my colleagues at the Institute for Geo- and Bioarchaeology (VU University) and the Amsterdam Archaeological Center (Universiteit van Amsterdam), my mum, dad and grandmother, and my friends who nodded so patiently when I told them about my latest results.

1

1 Introduction 1.1 A magical method? This study is about a method that has long since been used in archaeological prospection. For the last 30 years, magnetic methods have been frequently employed in both the archaeological academic and commercial world, on the one hand to map unknown archaeological sites and on the other to investi-gate known archaeological sites (see for example Clark 1996, Gaffney & Gater 2003, Linford 2006 for an overview). A good example of a magnetometer survey on a known archaeological site is the mapping of the site of the Standing Stones of Stenness on Orkney in the United Kingdom, which was one of the early magnetometer surveys with an archaeological purpose (Fig. 1). The plotted magneto-meter data clearly shows the partly buried ditch surrounding the megalithic monument, and a number of pits in the centre of the circle. On the northern edge of the monument lies a double linear anomaly that is, for the moment, interpreted as being post Neolithic ditches, as they appear to over cut the monument.

Figure 1 Magnetometer dataplot of the survey that was conducted by GSB Prospection at the Neolithic site the Standing Stones of Stenness, Orkney, United Kingdom. This is not the original dataset but the result of a repeated survey 25 years later. Extend 80 x 80 meter, range 0 (white) to 20 nT (black). Figure has been reproduced with kind permission. Another megalithic monument, probably the most famous of all, is the site of Stonehenge in Wiltshire in the United Kingdom. From an archaeologist’s point of view, not only the monument itself, but also the way that man has shaped the landscape around it is of major scientific interest. The proposed tunneling of the A303 road, running through Stonehenge's valley, and the development of a new visitor centre has sparked the need to investigate the area around the megalithic monument (Fig. 2).

2

Figure 2 The distribution of geophysical survey, almost all of it fluxgate gradiometer survey, in the Stonehenge area. Figure reproduced from David and Payne (1997) with kind permission. The method of choice was the magnetometer, and an area of 183 hectares was investigated (David & Payne 1997). This magnetic mapping on a landscape scale can be appreciated as a modern approach to an established method. As the focus of archaeology changes from the archaeological site to the archaeological landscape, the way that archaeological methods are employed changes accordingly. In the survey many features that were previously known to exist from old maps or from aerial photo-graphs were rediscovered and some new elements were added (see for example Fig. 3 for the results in the area directly around the monument). Most importantly, it is now much clearer what damage - if any - the infrastructural works are going to do to the monument in an entirely non-destructive way. Not all archaeological features can be mapped magnetically, however, and on some archaeological sites there is a general lack of magnetic contrast between the archaeological deposits and the un-disturbed subsoil, making the site unsuitable for magnetic prospection. The site of Easingwold in the United Kingdom, where none of the features that were excavated after the magnetometer survey had caused a detectable magnetic response, is an example (Weston 2004). This study is about the application of magnetic methods in The Netherlands, where the physical evi-dence of past societies can not be compared to that of the British Isles. And which is geologically different from the sedimentary rocks of Orkney, and from the chalk plain of Wiltshire. But there are examples of ‘successful’ magnetometer surveys on archaeological sites that approach the Dutch situation very well. The early Mediaeval site of Haithabu in Northern Germany for example. It has several ‘typical Dutch’ qualities. It was built out of wooden posts on a sandy spur next to the sea. Part of it has been waterlogged for a considerable length of time. And the results of the magnetometer surveys are excellent for giving information about the street pattern and the remains of individual buildings that are still present underneath the topsoil. During the last 30 years, in a time when archaeological geophysics was a ‘hot topic’ in universities in Germany, Austria, France, Italy, the USA, Japan and the United Kingdom, Dutch universities have never embraced the subject. In commercial archaeology, less than 20 magnetometer surveys have been carried out. But were the universities right not to participate in the research into these new methods? Or has Dutch archaeology missed out and are there opportunities for the successful appli-cation of geophysical methods in the Dutch situation?

3

Figure 3 Greytone plot of the entire magnetometer survey of the Stonehenge ‘Triangle’. Figure reproduced from David and Payne (1997) with kind permission. 1.2 Aims and objectives The aim of this project is to assess the possibilities of the application of magnetic methods for map-ping and evaluating archaeological remains in The Netherlands. The objectives of the project are: • To assess the use of magnetic methods in The Netherlands for mapping archaeological features

and sites. • To assess the use of magnetic methods in The Netherlands for archaeological landscape pros-

pection. • To define areas within The Netherlands where magnetic methods can and cannot be used within

the framework of Archaeological Resource Management. • To develop a quick soil sample based method to predict whether a magnetometer survey can

provide information about archaeological remains. 1.3 A Dutch perspective The interpretation of magnetic data presupposes a local geological knowledge, and the possibilities for the application of magnetic methods depend largely on the underlying geology. It is sometimes difficult to distinguish magnetic signals caused by archaeological features from geological variations, and archaeological responses can in some cases be overshadowed by geologically caused magnetic variations. For example, the linear magnetic anomalies at the Standing Stones of Stenness, that were tentatively interpreted as post Neolithic ditches at the start of this introduction, have in fact been caused by a - pre Neolithic!- volcanic dyke. Geological magnetic variations are common in igneous geological settings, but they occur in karstic and sedimentary landscapes as well.

4

On the other hand, the properties of the soil matrix in which archaeological features are shaped and preserved partly determine the magnetic properties of these features and their fills. Magnetic suscepti-bility enhancement depends on the presence of iron in the soil matrix. Archaeological activities on a hypothetical soil devoid of iron would not cause magnetic susceptibility enhancement. Apart from the presence of iron, other variables like the grain size, organic matter content and the hydrological situation make a soil more or less suited for magnetic enhancement and subsequent magnetic prospection. Because of the close relation between the successful magnetic mapping of archaeological sites and background geology, the core of this thesis is organized around the Dutch geological regions. As the magnetic properties of the soil depend largely on the lithological and organic composition of the soil, as well as on depositional, soil formation and post depositional processes, The Netherlands has been divided into three sections based on depositional environment; estuarine, wind blown and fluvial deposits. By using the depositional environment as discriminating factor, sites that have a similar lithological composition, and have depositional processes, soil formation processes and post depositional processes in common can be discussed as a group. Twenty-nine archaeological sites which are spread over these three groups have been investigated for their suitability for a magneto-meter survey. Moreover, soil samples have been collected on most of these sites in order to assess why certain sites are and others are not suitable for magnetometer surveys, in other words why certain sites do and others do not have a magnetic contrast. Investigations of archaeological and non-archaeological deposits for their magnetic susceptibility and for other magnetic properties aim to make that this thesis is more than a collection of survey results. Along side this geological approach, a number of typically Dutch themes have been investigated. A group of drowned village sites in the southwestern part of the country has been researched because their buried structures are difficult to map with traditional prospection methods like hand augering and surface collection. Off-site structures, field systems for example, pose similar problems, and the possibilities for a magnetometer survey on field systems in the western part of The Netherlands have been investigated. Other scattered sites include a battlefield, an extensive iron production site and a series of small dwelling mounds. Moreover, it has been assessed if magnetic prospection can be an additional method to aid the problematic prospection under plaggen soils, and in areas where peat extraction has taken place. 1.4 This study This introduction forms the first chapter of this study. The second chapter describes the history and the framework of archaeological prospection in The Netherlands. Prospection plays a crucial role in the cycle of archaeological resource management (ARM), which is why there is much interest to improve and add to current prospection techniques, this study is one of the examples of this trend. This chapter will explain the importance of non-destructive techniques for ARM and will set out what the current role for geophysical methods is. Additionally, archaeological prospection will be position-ed in the current archaeological practice and it will be investigated how the quality of (commercial) prospection activities is protected in the Kwaliteitsnorm Nederlandse Archeologie1 (KNA), and how the Nationale Onderzoeksagenda Archeologie2 (NOaA) sets the agenda for the future with respect to the scientific research into archaeological prospection. An introduction to magnetic methods is given in Chapter 3. Relatively much attention is given to the principles that underlie magnetic prospection, because it may not be possible to follow the argument in later chapters without understanding these principles. Most importantly, without an understanding about the way that magnetic susceptibility is related to induced magnetic anomalies, and how induced and remanent anomalies differ, this thesis can not be read.

1 Quality Norm for Dutch Archaeology. 2 National Research Agenda Archaeology.

5

Chapter 4 describes the methodology of this study, and starts with the way the approach to site selec-tion has changed during the course of the work. The choices that have been made with respect to the field and the laboratory equipment and sampling strategy are justified in the remainder of this chapter. The data of all the archaeological sites that have been investigated can be found in Appendix I, the sites have been ordered according to the list shown in Figure 12 (Chapter 4). The core of the study is laid down in Chapter 5 and 6, where the results of all the individual surveys are synthesized by depositional environment, the estuarine deposits (Chapter 5), and the wind blown and fluvial deposits (Chapter 6). The investigations in these chapters work towards answering the question in which areas in The Netherlands magnetic methods can be successfully used within the framework of ARM and why. Chapter 7 aims to be a catalogue of examples of the magnetic responses of different types of archaeo-logical features that were encountered in this study. The aims and objectives of the current study are related to ARM on the one hand, the importance of which is set out in the second chapter, and to the development of methods for archaeological prospection in general and to the methodological research into magnetic methods on the other hand. The importance of this study has to be appreciated on these three levels. The core of the book is concerned mainly with the basic level, the methodological research. The relevance of the results to the two other levels is discussed in the discussion (Chapter 8) and the conclusion (Chapter 9).

6

7

2 Archaeological prospection in The Netherlands 2.1 Introduction Archaeology is the scientific study of the physical evidence of past societies. It is thought that the buildings, features, objects and the way people have used and altered the landscape is a reflection of the way that they perceived the world and of the society that they were part of. The physical evidence of the past consists of visible archaeological features, e.g. tumuli, of invisible but known archaeo-logical sites, but largely of unknown, buried archaeological remains. In this respect, Deeben et al. (2005) have introduced the useful concept of archaeological stock, which is illustrated in Figure 4. The original stock is the total of all the archaeological remains that have been deposited, at any moment in time, in or on top of the ground. It is a fictitious stock, that has never existed at one moment in time and which has been and is being altered by biotic, anthropogenic and abiotic post-depositional processes, for example by human alteration of the objects or features that were deposited earlier, or by excavation. Hence part of this original stock is lost, another part has been gained because it has been excavated and recorded. The actual archaeological stock is what is left of the original stock after taking off the lost and the gained stock. Part of the properties of the actual stock is known through earlier research, but it is the unknown part of the actual archaeological stock that is the main object of archaeological research. This part of the soil archive consists of unknown archaeo-logical remains on unknown locations. Knowledge about this unknown stock is usually generated through archaeological excavation. Archaeological resources, however, are finite, and the activity of archaeological excavation destroys its own object of interest, by transferring part of the unknown stock to the gained stock; archaeologists kill their informants (Flannery 1982).

Figure 4 The relation between the known and the lost archaeological stock, and the unknown stock, which is the main object of archaeological research. After the text of Deeben et al. (2005).

8

A small part of the known stock is therefore preserved in situ as scheduled archaeological monuments. The first Dutch Monuments Act which provided a legal framework for the protection of archaeological sites came into effect in 1961. Not only scheduled sites were protected, but part of the unknown stock as well, as from this moment it was prohibited to carry out excavations without permission. The Act was replaced in 1988, in the new Monuments and Historic Buildings Act, in which under water archaeology was included. Before 1991, only designated parties could obtain a licence for archaeological excavation; universities, councils, provinces and the state service. From 1991, commercial archaeological companies could apply for an excavation licence. With a growing number of parties, the need to implement a quality system increased. At present, the Quality Norm Dutch Archaeology (Kwaliteitsnorm Nederlandse Archeologie (KNA)) is used to control the quality of the archaeological work that is carried out by all parties. A new Archaeological Monuments Act has been established in September 2007 for the implementation of the guidelines of the Malta Con-vention (1992). In this European Convention on the protection of the archaeological heritage of Europe, it is agreed that funding should be reserved for archaeological work in major private of public development. Further, the unknown stock will be better protected as the care of the archaeological heritage is integrated into the urban and regional planning. The changes in the Monument Acts reflect a policy shift from protecting individual archaeological monuments as a representation of the past, to future oriented archaeological heritage management (AHM), or archaeological resource management (ARM). Choices that are made today, determine what is left of the archaeological ‘resource’ in the future. ARM is often visualized as a cyclical process (Fig. 5).

��

Figure 5 The cyclical process of Archaeological Resource Management. Archaeological prospection plays a key role in the assessment and valuation of archaeological resources. See Willems (1997, 2008). Figure 5 is a schematic representation of this process, of the way choices are being made about the future management of archaeological resources. Whether an archaeological site is preserved or exca-vated, for example, can depend on the information that has been collected during the assessment and the valuation process. In these two steps, knowledge is generated about the unknown archaeological stock. It is in this domain that archaeological prospection must be introduced. Archaeological prospection can be defined as the scientific study of the spatial component of the physical evidence of past societies, which has been collected using techniques other than excavation. It broadly poses the same questions as, for example, excavating archaeology, but the way the answers are generated is different.

9

Most of the techniques that are used in archaeological prospection are practically non-destructive, which is a prerequisite in the process of ARM. The choices that have to be made about archaeological remains have to be based on information about these sites or objects, but extracting the information while destroying the carrier, as would be the case during archaeological excavation, could create a paradox. The archaeological heritage can not be managed without investigation, but investigation may lead to destruction, and by destroying the object of interest, the resource needs no longer be managed. Deeben et al. (2005) see a similar paradox in the known archaeological stock: the more knowledge there is about this part of the stock, the more damaged it is. Non-destructive techniques are needed to close the cycle of ARM. In terms of archaeological stock, unknown stock can become known stock through archaeological prospection (Fig. 4). This chapter will introduce the history of archaeological prospection in The Netherlands and in an international framework. Special attention is given to the development of geophysical techniques and to past experiences in the use of magnetometry as an archaeological prospection method. The chapter ends with a description of the framework in which archaeological prospection is currently embedded in The Netherlands. 2.2 The history of archaeological prospection in The Netherlands Aerial photography and surface collection The first decades of archaeological prospection in The Netherlands were entirely dominated by the investigation of manifestations of the buried archaeological record on the surface, i.e. by aerial photography and surface collection. The fist aerial photograph of an archaeological site was taken in England in 1906 and depicted Stonehenge. A growing amount of air traffic in and after the First World War led to an increase in interest in archaeology from the air. In the late 1940s and early 1950s, Von Frijtag Drabbe published a number of papers (e.g. Von Frijtag Drabbe 1947) devoted to the use of aerial photography for Dutch archaeology. A detailed description of the history of aerial photography in The Netherlands falls outside the scope of this study and can be found in Sueur (2006). In the same period, Stiboka3 was undertaking large scale soil mapping projects, during which archaeological surface finds were collected. These first accidental surface collection surveys were soon followed by more scientific surveys that were undertaken to solve a specific archaeological problem. Example of these surface collection programs are the surveys of Texel and West-Friesland which were initiated by the ROB.4 In the 1960s, archaeology was embraced by not formally trained or amateur archaeologists, who initially concentrated on surface collection too. Aerial photography generally provides information about archaeological remains that are buried at shallow depth in the form of crop marks or shadow marks, for example the so called Celtic fields (Brongers 1976) or visible at the surface (in the case of soil marks), although the aerial photography section of the Universiteit Gent has demonstrated that crop marks can originate from archaeological features that are buried under up to one meter depth of soil (J. Bourgeois, pers. comm.). In general, however, archaeological remains that are buried under deep deposits, both natural and anthropogenic, as well as earlier phases of multi-phase sites can not be seen in aerial photographs. This means that part of the archaeological stock will always be invisible to this technique (Fig. 4), and, in fact, most crop mark features are only visible at certain times of the year. Surface collection has similar limitations, this prospection technique can provide information about superficial sites and about sites that are buried at shallow depth and are damaged by post depositional biotic, anthropogenic or abiotic processes. Moreover, the relationship between the manifestations of the archaeological record on the surface and the buried actual archaeological stock is unclear (Flannery 1976). Rather than interpreting for example surface material as being a reflection of sub-surface remains, both classes of evidence should be appreciated as different and complementary sets of data (Haselgrove 1985).

3 Stichting Bodem Kartering, Dutch Soil Survey Institute, now renamed Staring Centrum. 4 Rijksdienst voor Oudheidkundig Bodemonderzoek, the State Service for Archaeology, now renamed to Rijks-

dienst voor Archeologie, Cultuurlandschap en Monumenten (RACM), the State Service for Archaeology, Cultural Landscapes and Monuments.

10

In order to investigate the archaeological record more fully, a shift from surface oriented investi-gations to sub-surface prospection was required. Sub-surface techniques How to gain knowledge of the subsoil without excavation, in a non-destructive way? The first inci-dental geophysical surveys were carried out in The Netherlands in the 1970s. Van der Kley (1968, 1970) investigated two Roman castellae by means of an earth resistance survey. At the time of intro-duction, the rate of data collection of this technique was very slow, but these surveys proved to be the first Dutch examples of the successful and non-destructive investigation of structural remains. In order to map buried archaeological sites that could not be detected with the surface oriented techni-ques, nor with electrical methods, a hand auger was first introduced in the Assendelver Polder project in 1980 (Terkorn 1991). By using this equipment, which was originally developed for soil scientific purposes, buried archaeological deposits could be mapped, investigated and related to their geological and chronological environment. In 1985 RAAP5 came into existence, originally a non-profit organization connected to the Universiteit van Amsterdam, which concentrated entirely on archaeological prospection. RAAP has been the main institution concerned with prospective archaeology in The Netherlands in the past 20 years. Initially the surveys consisted of a combination of surface collection, aerial photography and hand augering, and by employing this combination of methods, many new archaeological sites were discovered. Another part of the unknown archaeology stock was being made visible. Earth resistance and magnetometer surveys were selectively employed, and research into the applicability of these techni-ques was carried out from within RAAP. Magnetometry Because of the subject of this study, this section is concerned with the results of earlier magnetometer surveys. The first set of magnetometer surveys in the Netherlands was collected by Anderson (1996/ unpublished) for RAAP. In her study, that consisted both of magnetometer surveys and earth resistance surveys, 13 archaeological sites were magnetically surveyed. The work was conducted within the constraints of a commercial company, and the choice of sites that were investigated was undoubtedly influenced by commercial constraints. Most of RAAP's assignments for commercial work were for the investigation of castle sites; four of Anderson's sites were castles. On these sites the magnetics are generally noisy due to ex-situ bricks and metal, and the fact that most of the castle sites now lie in an urban environment. In two cases (Entinge and Merckenburg) walls, foundations and a well can be recognized in the magnetic data. On two furnace sites, a set of pottery furnaces (Holdeurn) and a set of roof tile furnaces (Swalmen) structures could be mapped in great detail due to the thermoremanent nature of the objects under investigation. Anderson’s magnetic datasets are generally dominated by objects and features with a remanent magnetization. The excavation Canisiuscollege in Nijmegen was situated in an urban setting, which, again, created a lot of magnetic noise. Some objects, stone, metal and a volcanic millstone caused magnetic anomalies due to their remanent magnetic properties. Much quieter is the data of the magnetometer survey on three prehistoric sites on wind blown sand deposits in the eastern and southern part of The Netherlands. A suspected hearth on a flint site (Borkeld and Elsenerveld) could not be mapped, and the features on two prehistoric burial sites (the urnfield of Someren and the tumuli group of Weert Boshoverheide) did not create any detectable magnetic anomalies. The Roman settlement on Kops Plateau, a terminal moraine, was first surveyed and excavated after-wards, but none of the magnetic anomalies seemed to correspond to the features that were excavated, but rather to the more recent paths and ditches. In Gennep, however, pits, ditches and an oven could be seen in the magnetometer data. This 4th/5th century site was located on deposits of the river Meuse, and it is the only example in Anderson’s study where non-remanent archaeological features have a detectable magnetic anomaly due to an induced magnetization. After Anderson’s study, a small number of magnetometer surveys has been conducted by RAAP, De Steekproef and DW Consulting, most of them on historical sites.

5 Originally an abbreviation for Regionaal Archeologisch Archiverings Project, Regional Archaeological

Archiving Project. Now renamed RAAP Archeologisch Adviesbureau, RAAP Archaeological Consultancy.

11

A brick church in Sexbierum in a matrix of clay caused a strong anomaly. In Neede Borculo the moat surrounding a small stronghold could be mapped because it was filled with brick from the demolished building. In Baaium the foundations of a church, built with tufa, caused a detectable magnetic ano-maly. An exception to the anomalies caused by remanent magnetization was the survey of a Michels-berg culture settlement (Schelsberg, Heerlen), here, a substantial ditch caused a negative magnetic anomaly. In 1998, one of the Neolithic farms in the experimental archaeology centre Archeon caught fire and burned down. A magnetometer survey was conducted over the remains before carefully excavating the features that remained after the fire. Unfortunately, the results of the survey were dominated by magnetic anomalies that were caused by pieces of metal building material (Orbons 1998). Hand augering Gradually, because of its success rate, hand augering became the preferred archaeological prospection method in The Netherlands, and not only in those areas where the surface oriented methods failed, i.e. the areas where archaeological sites were likely to be covered with Holocene deposits, but also in areas where surface collection and aerial photography had been successful, on the Pleistocene deposits in the east and the south. Groenewoudt (1996) investigated the instrumental variables that influence the discovery of an archaeological site by hand augering; grid spacing, type and size of auger and developed a triangular grid for hand augering. Almost ten years later, RAAP Archeologisch Adviesbureau, still the largest, but no longer the only archaeological prospector in The Netherlands, has continued the research of Groenewoudt, now including research into the effectiveness of mecha-nical drills (Tol et al. 2004). The increasing interest in the use of mechanical augers is also reflected a further report on the subject by the Universiteit van Amsterdam (Hissel & Van Londen 1994) on the qualitative comparison of mechanical and manual augering. During the 1980s and 1990s the methodological focus in Dutch archaeological prospection had shifted to hand augering, with relatively limited attention for established techniques like aerial photography or novel techniques like geophysical methods. An exception was the work of De Vries-Metz (1993) who investigated the application of remote sensing for mapping a prehistoric landscape in West-Friesland. Not only did she use novel remote sensing techniques, but she also propagated the use of remote sensing in ARM for the monitoring of scheduled archaeological monuments. Expanding the suite of prospection methods In recent years, the attention and funding for research into novel prospection techniques in Dutch archaeology has increased. This study into the application of magnetic methods in Dutch ARM is an example of the renewed interest in the methodological development of archaeological prospection, which has developed from the notion that prospection plays such an important role in the cycle of ARM. Alongside this project, there are two other lines of research in which the possibilities for chemical prospection (Oonk 2006) and the structural use of remote sensing (Sueur 2006) in Dutch archaeological prospection (and ARM) are being investigated. The first focus of the chemical pros-pection project is the development of novel extraction techniques for soil phosphate samples. Whereas soil phosphate determination as an archaeological prospection technique has been in use for a long time, better extraction techniques can increase the data quality and possibly refine the archaeological interpretation of phosphate distributions in the soil. A second branch of Oonk's research entails the investigation of the use of proteinaceous biomarkers in archaeological prospection, a topic which needs further research before it can be integrated into the current suite of prospection methods. Sueur's work has shown that there is a good potential for the use of remote sensing techniques in ARM in The Netherlands. There appear to be only few technical reservations, but the implementation of the techniques in policy and guidelines, most noticeably in the Quality Norm Dutch Archaeology (KNA), is lacking. One of the novel techniques that are advocated in his study is LIDAR. In the Technologie & Samenleving (Technology & Society) program for the innovative use of existing techniques and datasets, the Dutch Ministry of Economic Affairs for has funded a project on the archaeological use of the LIDAR dataset (Waldus & Van der Velde 2006), that was collected over The Netherlands for other purposes.

12

Two geophysical projects received funding in this same program, a project for the development of a portable sensor for in situ sediment characterization by measuring the concentration of �-ray emitting radionuclides (van Wijngaarden et al. 2002) and a project for the adaptation of the phased array technique that is used for telescopes to use in archaeological subsoil prospection. For the non-destructive prospection of archaeological remains in the water bottom (IMAGO 2003), for example ship wrecks, Rijkswaterstaat has conducted a comprehensive study using acoustic sub-bottom profiling, and ground penetrating radar (GPR) techniques. 2.3 The history of archaeological prospection from an international perspective Whereas Dutch prospective archaeology moved from the surface to the subsurface by focusing on hand augering, archaeological prospection in the United Kingdom and other European countries developed in a different way. Here, too, the first activities in the field of archaeological prospection were employed in aerial photography (e.g. Crawford 1928) and surface collection (e.g. Jenness 1930), but soon geophysical techniques were introduced for subsurface investigations in archaeological prospection. The earth resistance survey of Dorchester is generally considered to be the first scientific geophysical survey for archaeological purposes (Atkinson 1953). The success of this survey encou-raged archaeological institutions outside the United Kingdom to start using and developing electrical methods. This was soon followed by the development of the first magnetometer that was suitable for archaeological prospection in the 1950s. The development of a lightweight, high resolution magneto-meter created the opportunity to map large areas in limited time. At this point in time, archaeological geophysics was embedded into academic and public institutions in the United Kingdom,6 the United States,7 Italy,8 France9 and Germany10. In the 1960s the scientific community started to publish papers on the subject of geophysical prospection in archaeology (e.g. Aitken 1961, Lerici 1961, Black & Johnston 1962, Decker & Scollar 1962, Hesse 1962), first in journals dedicated to archaeology and archaeological sciences, from 1966 in the Italian periodical Prospezioni Archeo-logiche, and currently in Archaeological Prospection. Dutch archaeologists did not participate in this explosion of interest in archaeological geophysics. Although the first two geophysical surveys, which were carried in 1968 and 1970, were successful, geophysical methods did not get embedded in the research of academic or governmental institutions. In most European countries and the United States, geophysical techniques are and have often been employed alongside superficial prospection techniques like aerial photography and surface collection. The methodology focuses on the discovery and mapping of archaeological remains by superficial techniques, and the investigation and evaluation of the subsurface component of these remains by means of geophysical methods. Satellite imagery is the most recent addition to conventional aerial photography. Satellite images can be taken in the visual light or in any other electromagnetic spectrum, of which infrared is the most used for archaeological purposes. In the United Kingdom, Germany, Austria, Ireland and the United States the combination of remote sensing (including conventional aerial photography) and geophysical techniques has become a standard methodology for the archaeological prospection of ‘blank’ areas prior to development and in academia. An excellent example from Britain is the work that has been carried out in the Salisbury Plain Training Area (McOmish et al. 2002). In Germany, the Bayerischen Landesamtes für Denkmalpflege has surveyed extensive areas with a combined methodology and mapped many new archaeological sites (Becker et al. 1996), whereas in Austria, the Vienna Institute for Archaeological Science has concentrated on the development of methods to combine the data of aerial photography and geophysical surveys (Doneus et al. 1997).

6 The Ancient Monuments Laboratory and the Research Laboratory of the University of Oxford. 7 Museum Applied Science Center for Archaeology (MASCA), Pennsylvania. 8 The Lerici Foundation, Rome. 9 Centre National de la Recherche Scientifique (CNRS), Garchy. 10 Rheinisches Landesmuseum, Bonn.

13

Commercial companies started to carry out geophysical surveys for archaeological prospection after 1980, when the methodological development of the equipment allowed faster and more accurate surveys than in the first decennia of geophysical prospection for archaeology. This happened at a time when the amount of large scale, potentially disturbing infrastructural and housing projects was on the increase. Geophysical, and primarily magnetometer surveys are important tools for archaeological prospection in Archaeological Resource Management in many European countries and the United States. In the United Kingdom alone, the approximately 100 persons employed in archaeological geophysics have conducted hundreds of geophysical surveys. A new suite of methods, the electromagnetic methods, most notably the ground penetrating radar (Conyers & Goodman 1997, Leckebusch 2003) were added to the toolbox of archaeological pros-pection. Outside of The Netherlands, there are very few examples of coring being used as an archaeological prospection method (examples are Dockrill & Gater 1992 and Bates & Bates 2000), although a few early studies on shovel test sampling, which is comparable to hand augering, were carried out in the United States (e.g. Krakker et al. 1983, Kintigh 1988). During the years that were discussed in this paragraph, the focus of archaeological research gradually changed from the archaeological site to the archaeological landscape. In 1987, Heron and Gaffney announced that archaeological geophysics should change the object of its focus accordingly, ‘Archaeologists do not follow walls these days, so why should archaeogeophysicists?’ (Heron & Gaffney 1987:.78). This development is on-going, and at present individual archaeological sites are often investigated within their wider geological and archaeological framework. Still, research pro-grammes often lack the integration of a geophysical component (Gaffney et al. 1998). 2.4 Archaeological prospection in The Netherlands At the time of writing, archaeological prospection in The Netherlands has recently become embedded in the Quality Norm Archaeology (KNA)11 and the National Research Agenda (NOaA).12 As a result, companies, universities and other parties that carry out archaeological prospection work have to comply with a set of rules (the KNA) in order to ensure a minimal quality of the research. Most activities in the field of archaeological prospection are carried out in the cycle of archaeological resource management (ARM) (Fig. 5), and rarely without the urgency of the possible destruction of archaeological remains that may be present. In the field, archaeological work within ARM can be characterized by three phases: - reconnaissance, extensive, usually large scale, - mapping, semi-intensive, on the scale of groups of archaeological sites, - evaluation, intensive, on the scale of individual archaeological sites. Reconnaissance in The Netherlands is the domain of widely spaced (usually 40 x 50 meter) hand augering surveys and pick up surveys, whereas the second phase, mapping, is also conducted by hand auger, but on a more narrowly spaced grid (usually 20 x 25 meter). The most detailed phase of archaeological prospection is the evaluation, in this phase test trenching, narrow grid hand augering13 and geophysical methods are the techniques that are used to generate archaeological information. Whereas in other European countries geophysical methods are employed in all three phases of archaeological prospection, in the Netherlands geophysical techniques are only used in the evaluation phase, and rarely so.

11 Quality Norm for Dutch Archaeology. This text refers to KNA version 3.1. See www.sikb.nl for the full text. 12 National Research Agenda Archaeology, published on the internet only by the RACM. See www.noaa.nl for

currently available chapters. 13 Test trenching and hand augering are both semi-destructive methods. Some definitions of archaeological pros-

pection only include non-destructive methods, but Dutch archaeological practice has a focus on hand augering and to a lesser extend test trenching as prospection methods. The choice of the definition of prospection that was used at the start of this chapter, in which excavation is the only archaeological activity that is excluded, was directed by this focus on semi-destructive methods.

14

This discrepancy can possibly be explained by three factors; the historical development of archaeo-logical prospection in The Netherlands as was outlined above, the lack of interest in geophysical methods from academia and governmental institutions and the absence of a clear place for archaeological geophysics in daily archaeological resource management work. Only recently, archaeological geophysics has obtained a place in one of the universities that teach archaeology,14 but not in the archaeology department. Only for the small group of students that attend the Geoarchaeology course at this university geophysical techniques are part of the curriculum. In other archaeology courses, students are not educated in geophysical techniques for archaeology, neither is there any research in this subject. The State Service for Archaeology, Culture and Monu-ments (the RACM) lacks a section for archaeological prospection in general, or more specifically for geophysical prospection. The establishment of the Kwaliteitsnorm Nederlandse Archeologie has reinforced the dominancy of hand augering and test trenching over alternative prospection methods. In the reconnaisance phase, the latter two are the only recommended methods. In the two later phases, according to the KNA, it is possible to use geophysical methods. In practice, the prospection method that will be used in a certain project is laid down in a Statement of Requirement (Programma van Eisen (PvE)) and whether or not geophysical methods are included depends on the individual. A general lack of university education on the subject in combination with a lack of experience with and exposure to geophysical methods has led to the virtual absence of geophysical methods in recent Statements of Requirement. Another framework defining document is being made available, the Nationale Onderzoeksagenda Archeologie. The aim of this agenda is to define the framework for the questions that are being asked in the Statements of Requirement, and to form the scientific base for the day-to-day (commercial) archaeological work. The chapter about archaeological prospection is mainly methodological with relatively much attention for geophysical methods. Although addressing these questions is important, the type of methodological questions that are being posed will not aid in integrating geophysical techniques into daily archaeological practice. If the rather large document of the NOaA will, in practice be used for Statements of Requirement remains to be seen. 2.5 Conclusion • The different sections of physical evidence of the past can be represented as different types of

archaeological stock. • The unknown stock is the object of interest for archaeology. • Archaeological prospection is the study of the spatial component of the unknown stock with non-

or semi-destructive methods. Non-destructive techniques are needed in the cycle of ARM. • In The Netherlands, archaeological prospection developed from aerial photography and surface

collection to sub-surface prospection in the form of hand augering and limited application of geophysical techniques and remote sensing.

• Other European countries and the United States developed differently and focused on geophysical techniques for subsurface testing. The scale of geophysical investigations has changed from site to landscape.

• This discrepancy can be explained by the historical development of archaeological prospection, by the lack of academic and governmental interest in geophysical techniques and by the absence of a well-defined place for archaeological geophysics in the framework of ARM.

14 At the Institute for Geo- and Bioarchaeology in the Faculty of Earth and Life Sciences of VU University,

Amsterdam.

15

3 The principles of the application of magnetic methods for archaeo-logical prospection focused on The Netherlands

3.1 Introduction This chapter investigates the principles that underlie the application of magnetic methods in archaeo-logical prospection. Archaeological features can magnetically be mapped if they differ, in a magnetic sense, from the matrix that they are embedded in. This magnetic contrast is either of an induced or of a remanent nature. In order to explain the difference between induced and remanent magnetic anoma-lies, this chapter starts with listing the different types of magnetism. Then, magnetic susceptibility will be introduced and the way that the soil magnetic susceptibility is defined by the iron compounds that are contained in the soil. A focus shift to the archaeological component of soil magnetism will occur in paragraph 3.5, where the different pathways for the enhancement of soil magnetic susceptibility are discussed. If archaeological deposits are magnetically enhanced, induced magnetic anomalies may occur that can, under certain circumstances, be measured on the surface, whereas other processes may give rise to remanent magnetic anomalies. Paragraph 3.9 is concerned with the interpretation of these anomalies based on their shape, size and orientation. In the soil, dynamic processes are taking place that can cause changes in the type and in the amount of the iron oxides that are contained in the soil, and with it in the magnetic susceptibility of the soil. The most important of these soil processes are discussed in paragraph 3.10. 3.2 Different types of magnetism On an atomic level, magnetism is caused by the motion of electrons. In the traditional atomic model, electrons orbit the atomic nucleus, while spinning around their own axis. As electrons have an electric charge, the movement of the electrons creates a magnetic moment. Diamagnetism is caused by the orbiting motion of the electron. All materials display magnetic behavior in the presence of an applied magnetic field. Diamagnetism is a slight repulsive reaction that all matter has when it is subjected to a magnetic field; it obtains a small magnetic moment opposite to the direction of the applied field. Diamagnetism is the most common type of magnetism, but it is usually very weak and of the opposite sign to other types of magnetism, thus in the total magnetic moment it is often overshadowed by these other types of magnetism. The magnetic susceptibility of diamagnetic materials is weakly negative. Two of the main soil components are diamagnetic, i.e. quartz and water. Diamagnetic material can be recognized because of its linear dependency between an applied field and the magnetization of the material. All the other types of magnetism are caused by the motion of electron spin. In ferromagnetic mate-rials, for example, all electron spins are aligned in the same direction and interact with each other, thus creating a net magnetic moment. Ferromagnetic materials, most noticeably iron, are usually very crystalline and have a high magnetic susceptibility. They can posses a magnetic moment without an applied field, a property which is known as magnetic remanence. In antiferromagnetic materials all the electron spins are aligned in two opposite directions. The resul-ting net magnetic moment is zero. Antiferromagnetic materials, e.g. chromium, have a small positive magnetic susceptibility. Spin-canting, i.e. a small rotation of the spin directions, causes the two spin directions not to be perfectly opposite, this results in a weak magnetic moment, hematite is an example of a spin-canted mineral.

16

In ferrimagnetic materials, the electron spins are aligned like in antiferromagnetic material in oppo-site directions, but the two directions of magnetization have a different magnitude, which results in a net magnetic moment. This type of magnetism is similar to antiferromagnetism, but with a two-third rather than half of the spins aligning in one direction. Magnetite and maghemite are ferrimagnetic minerals that have a high magnetic susceptibility and lack a linear dependency to temperature increase. Ferro- and ferrimagnetic materials can obtain a magnetic remanence at room temperature, which is important for archaeological prospection, but above a certain temperature TC (Curie tempe-rature) these materials will display paramagnetic behavior. The Curie temperature of magnetite and maghemite, the most important ferrimagnetic iron oxides in the soil, is much higher than room tempe-rature (magnetite TC = 577 °C, maghemite TC = 547-713 °C). Antiferromagnetic materials will become paramagnetic above the Néel temperature (TN), e.g. for goethite 127 °C, and for lepidocrocite -196 °C. In paramagnetic materials, like aluminium, the spin of the unpaired electrons is randomly oriented, causing a number of magnetic moments in random directions with no net magnetic moment. Under the influence of an applied field, however, all these magnetic moments align to one direction. The magnetic susceptibility of paramagnetic materials is weakly positive, and there is a linear dependency between the applied field and the resulting magnetic moment in the material. Paramagnetism is inversely proportional to temperature. Lepidocrocite and ferrihydrite are paramagnetic at room tempe-rature. 3.3 Magnetic susceptibility Magnetic susceptibility is a measure for the ease with which a material can be magnetized in an applied field.

� = J / H where � is volume magnetic susceptibility J is magnetic moment of the material H is the intensity of magnetization (i.e. the magnetic moment per volume) mass magnetic susceptibility � = �/� where � is the density of the material

Every material has a different magnetic reaction to the application of an external magnetic field, and for soils this is no different. Soil magnetic susceptibility mainly depends on the mixture of iron compounds that is present in the matrix, further, it is influenced by the grain size, grain shape and the concentration of magnetic minerals in the soil. The most common soil iron oxides and their magnetic susceptibilities are listed in Table 1 and discussed in paragraph 3.4. Table 1 Properties of the most common iron oxides in Dutch soils. After Cornell & Schwertmann (2003). iron oxide magnetic susceptibility x 10-8 m3/kg magnetite Fe3O4 500001, 565002, 20000-1100003 ferromagnetic maghemite � Fe2O3 400001, 260002, 40000-500003 ferromagnetic hematite � Fe2O3 601, 402, 10-7603 spin-canted antiferromagnetic goethite � FeOOH 701,2, 26-2803 antiferromagnetic lepidocrocite � FeOOH 701,2, 40-703 paramagnetic* ferrihydrite 5Fe2O3·9H2O 402 paramagnetic* Data from: 1 Thompson & Oldfield (1986), 2 Maher (1998), 3 Hunt (1995); * At room temperature (> Néel temperature). For archaeological purposes it is important to realize at this point that all deposits, natural or anthro-pogenic, contain a specific combination of iron compounds, giving rise to a specific magnetic sus-ceptibility. Every deposit is unique, and is likely to have a magnetic susceptibility that is different from the deposits surrounding it. However, archaeological deposits and topsoil material often have an increased magnetic susceptibility compared to the undisturbed subsoil. The reason for this enhance-ment is the topic of paragraph 3.5.

17

3.4 Soil iron oxides In this section the iron oxides that are typical for temperate climate zones are summarised, using data from Taylor (1980), Weston (1999), Cornell and Schwertmann (2003) and Hansel et al. (2005). The chemical formula, colour and possible formation pathways of each iron oxide are briefly discussed, in addition to a description of well documented formation and transformation processes. However, the (trans)formation of iron in the soil is a complex process, which depends on many variables, including climate, soil pH, organic matter content and redox status of the soil. More advanced processes, not discussed here, are also possible and are best understood in the laboratory. In some cases the chemical formula for the iron oxides is the same (for example for hematite and maghemite) but the crystal structure is different, this is indicated with a prefix of � for the hexagonal and � for the cubic crystal structure. Goethite � FeOOH Goethite is the most common iron oxide in cool to temperate humid climates, where it usually coexists with lepidocrocite and ferrihydrite. It is yellowish brow in colour (7.5 to 10 YR). It can be formed from solid Fe(II) compounds like iron-carbonates or iron-sulphides, or reduced from Fe(III) by microbial action. Alternatively it can be transformed from ferrihydrite, or from ferric hydroxide (Fe(OH)3), which is a precursor to the more crystalline iron oxides, and can be formed after the water logging of a soil. Hematite � Fe2O3 Hematite is the second most common soil iron oxide, it mainly occurs in warmer to subtropical / tropical climates. It has a distinctively red colour (5YR to 10R). Hematite and goethite are closely related and can coexist, formation of either of the two iron oxides depends on temperature and drainage. Ferrihydrite, for example, can transform to goethite, but also to hematite in warmer and drier climatic conditions where the formation of hematite is preferred over the formation of goethite. In gley soils ferrihydrite may be the precursor of hematite. Lepidocrocite � FeOOH Lepidocrocite has been identified in different climatic zones, but not in calcareous soils. It is a very common iron oxide, usually orange in colour (5YR to 7.5YR). It is formed in seasonally wetting and drying (reductomorphic) environments, for example in iron pans. It can be transformed from ferrihydrite, or from magnetite after dissolution and consequent oxidation. Ferrihydrite 5Fe2O3•9H2O 5Fe2O3•9H2O is a possible formula for ferrihydrite, but other formulas have been proposed. Ferrihydrite is a hydrated ferric oxide, which can only be found in young (Holocene) deposits where its transformation to goethite or hematite has been impeded or delayed, or where circumstances are detrimental to the formation of more crystalline iron oxides like goethite. These are environments where there is sufficient Fe(II) oxidation in the presence of organic matter and silicate. Ferrihydrite occurs in gley soils, on the oxidizing / reducing boundary of the soil and in the B-horizon of podzol soils. Magnetite Fe3O4 Lithogenic magnetite occurs commonly in the coarse fraction of the soil. The abun-dance of magnetite in the topsoil when compared to the subsoil suggests a pedogenic formation (see § 3.5). Magnetite is a reduced iron oxide, which may oxidize to form maghemite, turning in colour from black to brown. In an oxidizing environment, magnetite will usually oxidize partly, the resultant iron oxide will belong to the magnetite / maghemite series, but will neither be purely maghemite, nor purely magnetite. Magnetite can be formed trough the reduction of hematite and goethite, or the reaction of ferrihydrite with Fe(II). Maghemite � Fe2O3 Maghemite is found mainly in tropical and subtropical regions, with localized deposits in temperate regions. The iron oxide occurs in concretions, or can be dispersed through the soil. It can be formed through the oxidation of magnetite, or the dehydration of lepidocrocite at a temperature of ~250 ºC. Taylor (1980) has synthesized maghemite from green rust during laboratory experiments, in a process that approximated the oxidation of a gley soil.

18

The iron mineralogy of the oxidizing part of calcareous estuarine, fluvial and marine deposits in Dutch subsoils is expected to be dominated by ferrihydrite and goethite, the former of which can, over time, transform into the latter. These two dominant iron compounds have a very low magnetic susceptibility (Table 1). Hematite, on the other hand, is unlikely to be formed from ferrihydrite in our climatic zone, except possibly in gley soils. In topsoils the non-ferrimagnetic compounds are expected to be mixed with ferrimagnetic, and thus high magnetic susceptibility iron oxides of the magnetite / maghemite series. The main contribution to the magnetic susceptibility of the deposits will come from these ferrimagnetic minerals.

Figure 6 Schematic representation of a number of important transformation processes between the most com-mon soil iron oxides in the Dutch soil. In non-calcareous deposits, lepidocrocite is likely to be present along side ferrihydrite and goethite. Here too, the main contribution to the soils magnetic susceptibility comes from the ferrimagnetic iron compounds of the magnetite / maghemite series. The soil is a complex dynamic system, and changes in for example hydrological circumstances or the soil chemistry have an impact on its iron minera-logy. The magnetic susceptibility of the soil most noticeably influenced when non-ferrimagnetic iron minerals transform to ferrimagnetic minerals, or the other way around. A number of relevant trans-formation processes between the iron oxides that are discussed here is illustrated in Figure 6. 3.5 Enhancement of magnetic susceptibility Le Borgne was the first to notice that usually, topsoil material has a higher magnetic susceptibility than the subsoil (Le Borgne 1955). This, he found, is caused by a concentration of ferrimagnetic minerals, mainly in the clay fraction of the topsoil. He proposed two principles of magnetic suscepti-bility enhancement, ‘fermentation’ (a term that was later replaced by ‘pedological processes’) and burning, both of which rely on the reduction and subsequent oxidation of non-ferrimagnetic iron oxides in the soil in order to form ferrimagnetic minerals of the magnetite - maghemite series. The most quoted example is: reduction oxidation hematite magnetite maghemite Because of the predicted lack of hematite in the Dutch soils, this reaction may be of lesser importance to this study. Probably more relevant is the thermal alteration of for example lepidocrocite to mag-hemite to hematite (Dearing et al. 2001); see Figure 6.

19

Pedological processes Weston calls the term fermentation ‘something of a misnomer’, all of the processes that cause magnetic susceptibility enhancement in the soil, apart from burning, can in his opinion be described as pedological and edaphic (soil biological) processes (Weston 2002). Long after Le Borgne, indeed an example of purely pedological enhancement was published in a paper by Maher and Taylor, on the discovery of the formation of magnetite in soils that were known not to have any primary magnetite input (Maher & Taylor 1988). There is no reason to assume that this pedological neoformation of magnetite would be preferential either to archaeological deposits or to the matrix surrounding these deposits, which leaves this type of magnetic susceptibility enhancement of no direct relevance to archaeological prospection. Edaphic processes Edaphic processes can be divided into the intracellular and extracellular bacterial formation of magnetite on the one hand, and the bacterially mediated processes that result in the formation of ferrimagnetic iron compounds on the other hand. In the early 1990s, Fassbinder & Stanjek (1993) are critical about the trend in archaeological prospection to explain the enhanced magnetic susceptibility of topsoils solely by heating processes and the subsequent formation of magnetite or maghemite. Not all archaeological sites, however, they argue, have burning episodes. In their research, the authors have found magnetite in an unburned post, and in their paper they show that this magnetite probably is bacterially formed. The soil bacteria that are likely to be responsible are magnetotactic bacteria, which produce intracellular magnetite in chains (Fassbinder et al. 1990). Extracellular bacterial formation of magnetite has been shown to occur in anaerobic sediments (Lovley et al. 1987). The microorganism GS-15 (now renamed Geobacter metallireducens; Lovley et al. 1993) can reduce amorphic iron to magnetite during the oxidation of organic matter. Both intra-cellular and extracellular magnetite forming bacteria need organic matter as a nutrient. Differences in soil organic matter content may lead to the preferential formation of bacterial magnetite in organic deposits, and may in this respect be important for the magnetic differentiation between archaeological and non-archaeological deposits. Dearing et al. (1996, 2001) argue that most of the topsoil enhancement in the temperate climate of England has to be attributed to the bacterially mediated formation of magnetite (or maghemite after oxidation) from ferrihydrite. According to the authors, low concentrations of secondary ferrimagnetic minerals can be explained either by the lack of an Fe supply for the formation of ferrihydrite, or by non favourable conditions for the iron reducing bacteria, or by post formation processes. The optimal conditions for iron reducing bacteria, they observed, is a wet but free-draining soils containing micro-pores, for example a silty loam. Ferrihydrite is the key mineral in this process. For The Netherlands it is important to note the grain size dependency for this type of magnetic susceptibility enhancement. Heating Heating may not be the only principle that causes the magnetic susceptibility enhancement of the soil, it is however the most important for archaeological prospection. The basic underlying chemical process for the enhancement is that non-ferrimagnetic iron oxides in the soil-like antiferro-magnetic hematite and goethite- that are exposed to high temperatures under reducing circumstances will convert directly into ferrimagnetic maghemite, or to maghemite through a magnetite phase. The reducing circumstances are created through the combustion of organic matter that is present in the soil. This obviously is a simplification of the process. As Weston has shown in his laboratory experi-ments, it is not simply heating that causes magnetic susceptibility enhancement, but there is a tempe-rature bracket in which enhancement can occur, if other conditions, the presence of organic matter and a fine grain size for the onset of reducing circumstances, have been met (Weston 2002). If soil is heated to temperatures that are too low to ignite organic matter (and create reducing circumstances), iron oxides may dehydrate or dehydroxylate rather than be turned into ferrimagnetic iron oxides, but if the soil is heated to very high temperatures, the magnetic susceptibility may decrease (Weston 2004). This has been further investigated by Maki et al. (2006) who encountered archaeological hearths that produced a negative magnetic anomaly during a magnetometer survey. In a laboratory reconstruction they found the primary (lithogenic) magnetite oxidized to maghemite at high tempe-ratures, followed by an inversion to hematite. That the other conditions, the presence of organic matter and a fine grain size, are important as well has also been shown in laboratory experiments (Weston 2004).

20

The coarse mineral soils that lack organic matter and lack a finer soil fraction reached a lower total magnetic susceptibility than finer soils or soils with more organic material. The coarser the material, the higher the temperature needed to obtain the maximum susceptibility in the medium. Moreover, Weston (2004) has found in his experiments that the waterlogging of soil can prevent magnetic susceptibility enhancement during heating. His explanation is that on the one hand the heat cannot penetrate sufficiently into the soil if it is waterlogged, on the other hand it is difficult to obtain higher temperatures. 3.6 Magnetic susceptibility in The Netherlands In England, a large topsoil magnetic susceptibility survey has been conducted that covered the whole land surface in approximately 1200 samples (Dearing et al. 1996, 2001). In The Netherlands there is no such national database in which values of the soil magnetic susceptibility are stored. During this study, however, topsoil samples were collected on a series of locations, the data of which is digested in Table 2. This very limited amount of samples has been used to create a map with topsoil magnetic susceptibility (Fig. 7). An impression of the variation in the magnetic susceptibilities in subsoil samples is given in Table 3. Table 2 The magnetic susceptibility of the topsoil samples that were collected during this study. N is the number of samples. site N range x 10-8 m3/kg mean x 10-8 m3/kg Aarle Rixtel 1 - 30.3 Beugen Zuid 13 19.87 - 39.58 30.02 Borgharen 4 33.85 - 44.79 40 Breda 5 0.92 - 16.17 10.46 Broekpolder 22 8.18 - 21.90 16.25 Deil 3 7.48 - 11.96 9.64 Den Dolder 3 6.36 - 9.38 7.7 Geldermalsen 3 8.92 - 9.25 9.1 Harnaschpolder 21 8.65 - 58.85 19.17 Heeten 21 13.12 - 39.19 22.43 Limmen 5 12.86 - 18.90 16.19 Meerssen 3 32.78 - 39.48 36.42 Midsland aan zee 1 - 0.76 Oostelbeers 1 - 12.81 Oostrum 4 8.88 - 14.61 11.21 Poeldijk 2 14.81-17.32 16.07 Raalte 1 - 8.53 Smokkelhoek 2 6.64 - 6.68 6.66 Stede Broec 3 10.16 - 31.10 18.25 Steenbergen 1 - 30.05 Stroe 3 1.11 - 1.31 1.22 Swalmen 2 36.82 - 55.98 46.4 Uitgeest 1 - 11.32 Vleuten 1 - 16.32 Wervershoof 3 7.85 - 8.80 8.23 Wijk bij Duurstede 3 129.73 - 240.48 178.03 Zaltbommel 5 16.70 - 27.97 20.29 Figure 7 does not pretend to provide an overview of Dutch topsoil magnetic susceptibility like its English counterparts does. It has been solely displayed because it seems to show a meaningful trend that may be important for the following chapters. Most strikingly, in the province of Limburg the topsoil magnetic susceptibility is elevated. This magnetically enhanced area possibly coincides with the extent of the presence of loess in the soil matrix. The area of enhancement is larger than the loess region, a phenomenon that may be caused by the transportation of loess sediment along the river Meuse. Elevated magnetic susceptibility of the soil can also been seen in the eastern part of The Netherlands. The two sample locations that cause the anomaly are metal working sites, however, and this type of high temperature activity can be expected to have caused an elevated soil magnetic susceptibility.

21

Figure 7 Topsoil magnetic susceptibility in The Netherlands from samples in this study, scale 8 x 10-8 m3/kg. Crosses indicate the sample locations. Data has been gridded using the minimum curvature routine, which gives the smoothest possible surface and is more suitable for small and scattered datasets than other interpolation methods like for example Kriging. On the other end of the spectrum, very low topsoil magnetic susceptibilities are recorded in Midsland and Den Dolder, both soils are coarse mineral soils, a quality that has been related to low magnetic susceptibility in paragraph 3.5. On the Brabant pleistocene sand plateau and in the fluvial area in the center of The Netherlands, magnetic susceptibility is usually relatively low and homogeneous. These observations will be further elaborated in Chapters 5 and 6. To add some detail, typical ‘magnetic susceptibility sections’ for the four prevalent Dutch geogenetic environments are displayed in Figure 8. As was discussed in paragraph 3.5, topsoil magnetic suscepti-bility is usually higher than in the layers underneath, a phenomenon that is observed in these sections. Exceptions are the estuarine examples of Smokkelhoek and Oostrum; both of these sections contain high magnetic susceptibility layers in the subsoil, and have relatively low magnetic susceptibility topsoils. The processes that underlie this phenomenon will be discussed in Chapter 5. Except for a high magnetic susceptibility topsoil, the other profiles have in common that their magnetic suscepti-bility decreases with depth. This may be linked to the general lack of organic matter at greater depth, which is needed for the bacterial or thermal enhancement of soil magnetic susceptibility. The Stroe profile is no exception; the two top layers are interpreted to be a recent addition and not part of the developed soil. The third layer is the actual high magnetic susceptibility topsoil layer.

22

Table 3 The magnetic susceptibility of the subsoil samples that were collected during this study. N is the number of samples. It must be noted that samples were taken across the whole undisturbed soil section. Different material properties have not been taken into account. site N range x 10-8 m3/kg mean x 10-8 m3/kg Beugen Zuid 14 0.77 - 18.74 4.80 Borgharen 6 5.65 - 11.81 10.45 Breda 5 0.12 - 1.02 0.44 Broekpolder 26 2.98 - 9.80 5.25 Deil 3 4.06 - 5.04 4.71 Den Dolder 8 0.53 - 3.52 2.23 Geldermalsen 4 1.41 - 6.98 4.07 Harnaschpolder 32 0.48 - 11.09 6.66 Heeten 21 4.48 - 108.51 23.84 Limmen 5 6.02 - 8.15 6.74 Meerssen 3 14.67 - 20.97 18.63 Oostrum 6 9.5 - 21.47 16.51 Poeldijk 5 0.73 - 24.79 9.66 Raalte 4 0.68 - 56.48 19.56 Smokkelhoek 44 -0.8 - 177.33 24.11 Stede Broec 6 3.05 - 6.02 4.34 Stroe 4 0.95 - 2.12 1.3 Swalmen 2 0 - 1.72 0.86 Uitgeest 4 1.74 - 5.32 3.13 Wervershoof 4 2.78 - 7.00 4.97 Wijk bij Duurstede 2 9.28 - 12.81 11.04 Zaltbommel 5 4.18 - 13.73 7.96 For archaeological prospection purposes, the two characteristics that were discussed, i.e. a high magnetic susceptibility topsoil and a decreasing magnetic susceptibility down profile, are of vital importance for a potential magnetic differentiation of the feature fill and the undisturbed matrix, and subsequently for the formation of induced magnetic anomalies. Negative archaeological features can get a positive magnetic expression if they are filled with a higher magnetic susceptibility fill. This is more easily obtained in a situation where topsoil and surface material have higher magnetic suscepti-bilities than the subsoil. The greater the differentiation between topsoil and subsoil was in the past, the more likely it is that archaeological features now cause an induced magnetic anomaly. Based on the presented data, there are hints as to where the application of magnetic methods for archaeological prospection in The Netherlands may have more or less potential. In the south eastern loess area and in the Meuse valley, for example, topsoil magnetic susceptibility is high, and differences between topsoil and subsoil are large, whereas in the northern areas the two layers seem to be far less differentiated. Based on this limited information, the loess area seems to have more potential for the use of magnetic methods in archaeological prospection than the marine clay areas. The aeolian deposits in the southern and eastern part of The Netherlands seem to be well differentiated, but the current situation, where plaggensoils have been added since the prehistory, is unlikely to correspond to the situation at which these soils may have been inhabited in the prehistory. The relation between the high magnetic susceptibility fill of archaeological features and magnetic anomalies is the topic of the following paragraphs. 3.7 Induced magnetization Induced magnetization is the magnetic moment that occurs in a material under the influence of an external magnetic field. It has a linear dependency on both the magnetic susceptibility of the material and on the applied magnetic field: Ji = �H

23

where Ji is the induced magnetization, � is the mass magnetic susceptibility and H is the field strength of the applied magnetic field. The direction of magnetization is usually identical to the direction of the applied field.

Figure 8 The magnetic susceptibility of eight typical soil profiles. Values on the right hand side of the cores represent the magnetic susceptibility x 10-8 m3/kg. For matters of clarity the soil descriptions have been simpli-fied to the three main classes of sand, clay and peat. Detailed soil descriptions of these profiles can be found in Appendix III.

24

The soil matrix, the object of archaeological magnetic prospection, is continuously magnetized by the earth’s magnetic field. Differences in soil magnetic susceptibility, under the influence of the earth’s magnetic field, are the cause of differences in induced magnetization in the soil. 3.8 Remanent magnetization Some objects or materials can be magnetized without the application of an external magnetic field, this is called remanent magnetization. An example are ferromagnetic materials, like iron. In the case of objects, the direction of magnetization is along the easy axis of magnetization in the object. Most relevant for archaeological prospection is thermoremanence, the remanent magnetization that an object or feature can obtain after is has been heated to high temperatures. The transition of non-ferrimagnetic minerals into ferrimagnetic minerals causes the magnetic susceptibility of the material to increase, but these ferrimagnetic minerals can also carry a remanent magnetization in zero external field. In the heating process the direction of the magnetic particles is disturbed, on cooling the (newly formed) magnetic particles will align along the direction of ambient magnetic field. In situ, the direction of magnetization will be approximately parallel to the direction of the earth’s magnetic field on cooling. Measurements of this direction, in combination with knowledge about the past declination and inclination of the earth’s magnetic field, can be used for magnetic dating. Ex situ, the magnetic moment of thermoremanent objects or features can have any direction. Examples of thermoremanent objects are bricks, tiles, and igneous rocks, features that have obtained a thermoremanent magneti-zation are, for example, hearths and kilns. There can be other reasons than heating and cooling why magnetic particles align themselves in one preferred direction. In a material that carries no magnetic remanence, but that does contain an amount of ferrimagnetic minerals, a mechanical shock may cause the magnetic minerals to align in one predominant direction (this is named shock or shear remanent magnetization). It has been suggested that mud bricks, which are made in moulds from which they are removed by hitting the ground with force, carry this type of remanent magnetization (Kattenberg 1999). Chemical remanent magnetization is laid down during the formation of magnetic minerals that grow in one predominant magnetic direction. Detrital remanent magnetization is a post-formation process that is caused by magnetic particles which lay themselves down in one magnetic direction after having been in suspension. These two types of remanent magnetization may not be relevant for the magnetic detection of archaeological features, but magnetic anomalies that are caused by chemical or detrital remanent magnetization may be encountered during a magnetometer survey for archaeological pur-poses. 3.9 Magnetic anomalies Variations in both induced and remanent magnetization in the soil influence the total ambient magnetic field, which is the combination of the earth’s magnetic field and other magnetic moments that are present near the earth’s surface. In a magnetometer survey, anomalies in the total ambient field may be meaningful for objects or features that are buried underneath the surface. Magnetometer instruments are constructed to measure (one of the components) of the ambient field (see paragraph 4.3.1). The main contribution will, in almost all cases, come from the earth’s magnetic field, followed by any remanent and induced magnetizations. 3.9.1 Direction of magnetization In most cases, the magnetic responses of features with an induced or a remanent magnetization can not be distinguished, because their direction of magnetization is similar. Features with a thermoremanent magnetization that have remained in the position where they cooled, will have a magnetization in the direction of the ambient field at that moment, which is mainly dominated by the earth’s magnetic field.

25

The direction of the earth’s magnetic field has changed through archaeological time, but on an archaeological timescale not so drastically that this shift in orientation can be seen in the magnetic anomalies that are mapped in a magnetometer survey. Magnetic anomalies that are caused by the induced magnetization of archaeological feature fills with a magnetic susceptibility contrast are also magnetized in the direction of the earth’s magnetic field, and cannot be distinguished from the anomalies that stem from a remanent magnetization. Objects with a remanent magnetization like pieces of brick or metal, however, may have any direction of magnetization and can, because of their direction, be easily distinguished in the results of a magnetometer survey. The size, shape and orientation of the magnetic anomaly that are caused by objects or features which either have an induced or a remanent magnetization can be used for the interpretation of magneto-meter data, which is discussed in the next paragraph. 3.9.2 Size and shape The size of a magnetic anomaly depends on the strength of the remanent or induced magnetic contrast, the volume of the object or feature causing the anomaly, and the depth at which it is buried. The properties of a feature or object causing a magnetic anomaly can be modeled through inversion based on its magnetic response, for example in the program MAG3D15. This software assumes that the anomalies that it analyzes are caused by magnetic susceptibility distributions in the soil, and induced magnetization under the influence of the earth’s magnetic field. A program for the inversion of metal objects is UXOLab16, which was developed for the analysis of the magnetic response of unexploded ordinance. In reverse fashion, an anomaly can be forward modeled by defining a set of relevant variables, magnetic susceptibility contrast, volume and depth of burial of an object or feature. In this study the program Modeller17 was used. Only magnetic susceptibility contrasts have been investigated, anomalies that were caused by magnetic remanence have not been modeled in this study. In Figure 9a, a typical induced magnetic anomaly, a pit, buried at 0.25 meter depth, has been dis-played as a magnetometer response at Dutch latitudes if traversed from north to south. The response is parabolic, with its peak slightly off-set to the south of the causative object and a negative dip to the north side of the peak. In Figure 9b the same pit is buried at greater depth, between 0.75 and 1.25 meter. Because of the greater distance between the feature and the magnetometer, the response is much weaker, but in size it is actually wider. The peak and the dip are less pronounced. In Figure 9c a smaller pit is modeled, that is buried at the same depth as the pit in a. Because of the decrease in volume, the response is weaker and narrower than that of the original pit in Figure 9a. Figure 10 is an example of the geometry of the peak and trough of a magnetic anomaly in a 2D view, this is the type of display that is the most commonly used graphical way to display magnetometer data. The data is a section of the survey that was carried out on the metal working site of Heeten Hordelman. Because the site was excavated after the magnetometer survey, anomaly A is known to represent a rubbish pit. The geometry of the magnetic anomaly, a peak with a relatively small halo shaped trough on its north side, is a typical example of induced magnetization. Object or feature B, on the other hand, was not recorded during the excavation. It consists of a central trough with a circle of higher values surrounding it, which is, because of its orientation, indicative for remanent magneti-zation. The anomaly was possibly caused by a piece of metal in the topsoil. Two small anomalies marked with C in Figure 10, are clear peak-and-trough anomalies. Because of their east-west orienta-tion, the features can be interpreted as having a remanent magnetization. If the objects causing the anomalies would have been turned by 90 degrees anti-clockwise, the anomalies could have been misinterpreted as features with an induced magnetization.

15 MAG3D; A Program Library for Forward Modelling and Inversion of Magnetic Data over 3D Structures.

Developed under the consortium research project Joint/Cooperative Inversion of Geophysical and Geological Data (JACI), UBC-Geophysical Inversion Facility, Department of Earth and Ocean Sciences, University of British Columbia, Vancouver, British Columbia.

16 UXOLab is a software package for analyzing magnetic and electromagnetic data for the purpose of detection and discrimination of unexploded ordnance (UXO).

17 Modeller version 2.11, © Nic Sheen 1995.

26

Figure 9 Modeled gradiometer res-ponses (vertical component). North is pointing to the right. a: typical archaeological pit with a high magnetic susceptibility fill; b: the same pit as in (a) buried at a greater depth; c: a smaller pit buried at the same depth as (a); d: a com-bination of graph (a), (b) and (c). Deeper burial of the same object (compare (a) and (b)) causes a weaker and broader magnetic ano-maly. Shrinking the object (compare (a) and (c)) causes a weaker and nar-rower anomaly. Note the negative trough is located on the north side of the anomaly.

27

Figure 10 An example of the response of magnetized objects or features in a magnetometer survey in a 2D display. Anomaly A represents the response of a rubbish pit with a high magnetic susceptibility fill. The positive peak and the slight negative halo on its north side are typical features of an induced magnetization. Anomaly B has a central trough surrounded by a positive halo, an indication of a remanent magnetization. The anomalies marked with C have a clear peak with a trough to the west. Because of this orientation they can be identified as remanent magnetic objects or features, if the trough has been oriented to the north, than the features could have been misinterpreted as being caused by induced magnetization. A 3D view of Figure 10 can be seen in Figure 11.

Figure 11 Trace plot of the data of Figure 10, looking west. The pit anomaly (A) is located on the foreground. Note the sharp trough in anomaly B. 3.10 Post depositional processes Remanent magnetization in an object or feature is usually firmly preserved in the material, but mechanical processes can break up and change the direction of their magnetic moment.

28

A remanently magnetized object with one predominant direction of magnetization, e.g. a kiln, can desintegrate into pieces that each still have a remanent magnetic moment, but all in a different direction. Chemical and detrital magnetization in sediments may be destructed if the soil material is agitated for example by flowing water. Magnetic susceptibility, and therefore induced magnetization, on the other hand, is much easier affected by post depositional processes. The processes that are most likely to influence the magnetic susceptibility of a soil are wetting and drying, because these processes can lead to the transformation or the movement of iron oxides (Fig. 6) in the soil. If a soil is continuously waterlogged for a long period of time, iron oxides can be reduced to hydrated iron hydroxides (Weston 2002). Further reduction will change iron in the ferric form (FeIII) to a ferrous form (FeII), which is accompanied by a soil colour change from red, yellow and brown to blue and grey. The ferrous iron is likely to be redistributed or leached from the soil profile. During long lasting waterlogging episodes, even ferrimagnetic minerals can dissolve, and they may be removed from the soil profile (or be transformed into ferric hydroxides; Weston 2002), causing a drop in soil magnetic susceptibility. The sequence of the solubility of iron oxides is ferrihydrite > lepidocrocite > maghemite > goethite > hematite, but the latter two iron oxides may be in reverse order (Cornel & Schwertmann 2003). Goethite and hematite are the most stable iron oxides, and are the end product of many transformation routes (see also Fig. 6). Transformations of the soil iron oxides caused by water-logging will influence the magnetic susceptibility of the soil. Changes in the iron mineralogy that may occur under the influence of sea water logging are discussed in Chapter 5. During the process of gleying Fe(III) is dissolved under long lasting reducing conditions and relocated to a higher level, to the zone in which the groundwater fluctuates, causing the typical iron staining. Mullins noticed that gleyed horizons have unusually low susceptibilities, which, he states, is caused by the dissolution (and movement) of ferrimagnetic maghemite under reducing circumstances (Mullins 1977). The oxidation of redeposited iron can possibly lead to the formation of a number of different iron oxides, most noticeably ferrihydrite, but, according to laboratory experiments (Taylor 1980) also for example hematite and maghemite. The process of leaching occurs, for example, during podzolisation. Iron is leached from the upper part of the soil and redeposited at a lower level, in the case of a podzol in an ‘iron pan’ in the form of (antiferromagnetic) lepidocrocite. Weston has seen that subdued magnetic susceptibilities occur in coarse mineral soils, especially in those soils that have a high water table, suggesting leaching of iron oxides to be the cause. In order to find out if this was the case, he set up two leaching experiments: - leaching with H2SO4, mimicking leaching by acid rain; - leaching with EDTA, mimicking leaching by plants and soil biota. The experiment showed coarse (sandy) soils are more easily leached of iron than finer soils or soils with a high organic matter content, and only very acidic conditions can mobilize iron (Weston 1999). Moreover, there was no noticeable influence on the magnetic susceptibility of previously ignited samples. The enhanced magnetic susceptibility in both clay and sand samples appeared to be more resistant to leaching than the in the non-ignited samples (Weston 2004). 3.11 Conclusion In this chapter the principles behind magnetic prospection in archaeology have been set out. The most important points, which are needed to understand the following chapters, are summarized here. Magnetic susceptibility • The magnetic susceptibility of a soil is mainly determined by the amount and type of soil iron

oxides, in The Netherlands magnetite, maghemite, lepidocrocite, goethite, hematite and ferri-hydrite are expected to be the most common iron oxides.

• Every (soil) material has a different magnetic susceptibility. • Topsoil usually has an enhanced magnetic susceptibility compared to the undisturbed subsoil, this

also appears to be the case in samples from The Netherlands.

29

• Archaeological deposits often have an enhanced magnetic susceptibility. • Magnetic susceptibility enhancement can be caused by heating (but is impeded by waterlogging

during heating) or by bacterial action, both processes depend on the presence of organic matter. • The magnetic susceptibility of finer soil fractions is more easily enhanced than coarser fractions,

in the Dutch soil samples coarse samples appear to have a lower magnetic susceptibility than the finer samples.

Magnetic anomalies • Magnetic susceptibility differentiation between the topsoil and the subsoil is a favorable circum-

stance for the formation of magnetic anomalies, in The Netherlands this differentiation appears to be best in the loess area and Meuse valley.

• Differences in magnetic susceptibility lead to induced magnetic anomalies. • Remanent magnetic anomalies can be caused by heating or other processes. • Magnetic anomalies can be interpreted based on their shape, size and orientation. Post depositional processes • Prolonged water logging, severe leaching and gleying can cause the dissolution and movement of

iron, and will usually lead to a change in soil magnetic susceptibility.

30

31

4 Methodology 4.1 Choice of sites At the start of this project, it was intended to investigate three large sites, or rather, site related land-scapes for this study. It was thought that the magnetometer survey in these three areas would identify archaeological features and possibly geological phenomena that could subsequently be sampled for laboratory investigations. In the first months of this project, two locations were investigated by means of a magnetometer survey combined with hand augering and soil sampling, the site of Broekpolder and the site of Harnaschpolder. In fact, very few archaeological features could be mapped in the magnetometer surveys in these areas, but geological features were abundant in the Harnaschpolder survey results. The reason why the archaeological features did not cause magnetic anomalies was investigated, as well as the nature of the geological anomalies in Harnaschpolder (see Chapter 5). The data collected at these initial two sites, however, did not appear to be sufficient for the overall project. The lack of identifiable archaeological features was the main reason to change the approach to site selection. On both of the sites, test trenches had been excavated before the magnetometer surveys only, which meant that archaeological features could not be sampled directly in excavation. Neither was it possible to sample the magnetically invisible features based on the magnetometer data. There was a need for better ground-truthing of the data. A practical reason for changing the approach was the difficulties with gaining access on a large set of fields with different owners. The new approach consisted of the selection of small areas that were being or were about to be excavated. This way, magnetically invisible archaeological features could be sampled in excavation, and a direct comparison between the magnetometer and the excavation data could be made. Criteria for site selection were: - preferably excavation of the site after the magnetometer survey, - archaeological features proven to be present or very likely to be present, - archaeological features expected to be present in the top meter of the soil matrix, - green field site. A negative aspect of this new approach was that site selection depended on the availability of sites that were about to be excavated. In Table 4 it can be seen that for ten of the 29 sites that were investi-gated, it was possible to carry out a magnetometer survey before the archaeological excavation took place. Apart from Broekpolder and Harnaschpolder, two more sites were surveyed after the excava-tion had been carried out. The remainder of magnetometer surveys was conducted on sites that were not intended to be excavated, mainly on scheduled archaeological monuments. These sites were selec-ted for different reasons, usually because they represent a type of site for which there were no on-going excavations at the time of research. This set includes four so-called drowned villages (Valke-nisse, Polre, Oostende and Steenbergen), a Roman road (Swalmen) and a Medieval road (Kolhorn) and one or possibly two sites with tumuli (Ossenisse) and an urnfield (Slabroek). The strategy for the magnetic susceptibility sample collection was to collect soil samples al all sites where a magneto-meter survey had been carried out. This could not always be achieved. Samples for magnetic suscepti-bility measurements have been collected from excavation trenches where possible. Some samples have been taken by hand auger, either as a complementary dataset to the samples that were collected directly from the archaeological features, or because this was the only opportunity to collect samples, as was the case on five locations. At the bottom of Table 5 five sites are listed where no magneto-meter surey took place, but which were available for sampling during excavation.

32

4.2 Inventory of sites In Chapter 3 the variation in background magnetic susceptibility in The Netherlands was discussed, and it could be seen that magnetic susceptibility depends mainly on the type and abundance of iron minerals in the soil. The outcome of any transformation processes in the iron mineralogy is also related to the initial iron mineralogy. Thus, the iron mineralogy influences the magnetic susceptibility of a soil, as well any potential changes in magnetic susceptibility (see also § 4.3.2.3 on heating experi-ments). In order to cover as many soils of a different iron mineral composition as possible, the archaeological sites that were investigated for this study were also selected on their location. The selection was limited by the availability of sites where archaeological work was being carried out. The locations of the sites that have been investigated in this study are displayed in Figure 12. In this study very little attention has been given to the northern provinces of Friesland, Groningen and Drenthe. The reason for this is twofold, first, there are less ground disturbing activities in this region, and hence less archaeological projects, this is especially true for the province of Drenthe. Secondly, the archaeo-logical remains on the sites that are excavated in Friesland and Groningen are usually not contained in the top meter of the soil matrix.

Figure 12 The location of the sites that were investigated in this study. The sites of Oostende (Belgium) and Spalding (United Kingdom) have not been displayed.

33

Table 4 List of archaeological sites that have been investigated in this study. The table indicates whether magnetometer surveys were conducted before or after archaeological investigations or if no excavations were carried out. Where possible, samples for magnetic susceptibility measurements were retrieved from archaeological excavations, in other cases from coring. site magnetometer

survey before excavation

magnetometer survey after excavation

magnetometer survey; no excavation

soil samples from excavation

soil samples from hand augering

Harnaschpolder • • • • Heeten • • • • Breda • • Limmen • • • Meersen • • Meteren • Poeldijk • • Spalding (UK) • • Wervershoof • • Zaltbommel • • Beugen Zuid • • • Broekpolder • • Borgharen • • Kolhorn • Oostende (BE) • Ossenisse • Polre • Slabroek • Smokkelhoek • • Stede Broec • Steenbergen • Swalmen • Uitgeesterbroek • • Valkenisse • Deil • Den Dolder • Raalte • Uitgeest • Wijk bij Duurstede • Apart from a geographical spread, the selection of sites needed to be representative of the archaeo-logical record - as far as this was possible - both in period and in type of site. Table 5 shows the dating for the sites that were investigated in this study. Where relevant, only the date of the features or of the section of the site that has been studied in the framework of this research is given. The dating of the actual archaeological site may be much wider. The ‘modern’ box is ticked if modern features have produced magnetic anomalies in the magnetometer data or if a modern feature has been sampled. The table shows that all archaeological periods, starting with the Bronze Age, are represented in the data-set. Sites dated before the Bronze Age have been excluded because these often lack the abundance of negative features like pits and ditches, that can be seen on the sites from later periods, and which are the targets for magnetometer surveys. In Table 6 the type of site for the archaeological sites under investigation has been listed. Sites have been divided into five groups. Almost all of the sites that have been investigated are settlements, which may reflect the focus of current archaeological research on settlement sites. Moreover, it is likely that the majority of the archaeological sites in The Netherlands falls into the category of settlement site, as a high percentage of the current archaeological monuments do (Lauwerier & Lotte 2002). Three sites are classified as off-site because of the field systems that were excavated here (Broekpolder and Harnaschpolder) or because of off-site activities (the battlesite of Breda). Burial sites that have been investigated date from the Bronze Age (Den Dolder, Slabroek and possibly Ossenisse) and the Middle Ages (Borgharen). Attention to the magnetic expression of industrial acti-vities has been given on sites for iron production (Heeten and Raalte), salt making (Spalding), peat extraction (Kolhorn and Smokkelhoek) and quarrying (Meerssen).

34

Table 5 Archaeological periods for the archaeological sites that have been investigated in this study. The ‘modern’ box is ticked only if modern features cause a magnetic anomaly in the data of the magnetometer survey or if a modern feature has been sampled. site Bronze Age Iron Age Roman Period Medieval and post Medieval modern Broekpolder • • • • Ossenisse • • Slabroek • • Deil • Den Dolder • • Stede Broec • • Wervershoof • Meteren • • Uitgeest • • Beugen Zuid • • Breda • • Borgharen • • • Harnaschpolder • • Heeten • Meersen • • Poeldijk • Raalte • Smokkelhoek • • Spalding (UK) • Swalmen • Uitgeesterbroek • Zaltbommel • Kolhorn • Limmen • Oostende (BE) • Polre • Steenbergen • Valkenisse • Wijk bij Duurstede • Table 6 Type of archaeological site for the sites that have been investigated in this study. site settlement off-site burial site industrial infrastructure Breda • • Broekpolder • • Harnaschpolder • • Borgharen • • Den Dolder • • Ossenisse • Slabroek • Heeten • • Meersen • • Smokkelhoek • • Spalding (UK) • • Raalte • Kolhorn • • Limmen • • Poeldijk • Swalmen • Beugen Zuid • Deil • Meteren • Oostende (BE) • Polre • Stede Broec • Steenbergen • Uitgeest • Uitgeesterbroek • Valkenisse • Wervershoof • Wijk bij Duurstede • Zaltbommel •

35

4.3 Instruments 4.3.1 Magnetometer survey Magnetometers are instruments that can measure the magnitude of a magnetic field. In archaeological prospection, the object of interest is the earth’s magnetic field. Magnetometers can either measure the magnitude of the total field, or of the horizontal or the vertical component of a magnetic field, or a combination of the horizontal and the vertical component. There are different sensors being used in ground based magnetometer surveys. For archaeological purposes, a very sensitive sensor and fast data acquisition are needed. All the magnetometer surveys that are presented in this study have been carried out with fluxgate gradiometer instruments. This type of magnetometer is widely used in archaeological prospection. The English Heritage guideline for geophysical surveys that was published in 1995 (David 1995) assumes that a magnetometer survey for archaeological purposes is conducted with a fluxgate gradio-meter, unless there are specific reasons to use another type of magnetometer. At present, however, an increasing amount of resonance magnetometers, especially caesium vapour magnetometers is used in archaeological prospection (Gafney & Gater 2003: 41). The sensitivity of these instruments is very high, up to 0.001 nT, which is 100 times more sensitive than a fluxgate sensor. Caesium magneto-meters are less portable and more expensive than fluxgate magnetometers, and problems arise in the measurement of steep magnetic gradients. Because of their weight, they are often placed on carts, especially in gradiometer configuration where one caesium vapour sensor is placed above another sensor or an array of sensors. This type of magnetometer has not been used in Dutch archaeological prospection yet. 4.3.2 Fluxgate magnetometer A fluxgate sensor typically consists of a pair of parallel mumetal cores each surrounded by a primary coil, these coils are wound in opposite directions to each other. A set of secondary coils is wound around the primary coils, again in opposite directions to each other and in respect to the primary coils. An AC current through the primary coil drives the mumetal cores in and out of saturation. The magnetic field that is thus produced causes a voltage in the secondary coils that is in-phase but in opposite direction to each other, the combination of the two voltages is always zero. If the sensor is now placed in a secondary field, for example the earth’s magnetic field, then a component of the external field will be parallel to the axes of the cores, resulting in an earlier saturation of this core, and a phase shift in the voltage. The combination between the two voltages is now non-zero, and this, being proportional to the field strength of the component of the magnetic field that is parallel to the axes of the mumetal cores, is the output of the fluxgate magnetometer. For a full description of the principle of the fluxgate sensor see Scollar (1990) or Reynolds (2002). A fluxgate sensor is only sensitive to the component of the external field that is parallel to its axis. In archaeological prospection the vertical component of the earth’s magnetic field is usually measured. Diurnal variations in the terrestrial magnetic field have magnitudes that are comparable to magnetic variations that can be caused by buried archaeological features. For this reason a gradiometer configu-ration is often preferred over a single sensor magnetometer. In a gradiometer two fluxgate sensors are placed a set distance apart, in line with each other in the sensitive direction. The sensor that is furthest away from the object of investigation, in this case the earth’s surface, is mainly influenced by the earth’s magnetic field and its variations. The sensor closest to the surface is influenced by both the terrestrial field and the magnetic variations in the soil matrix. Distracting the first output from the second produces a reading that is less influenced by diurnal variations. The advantage of fluxgate sensors is that they are light and portable, relatively cheap to manufacture, and that the sensors have no problems with large gradients in the magnetic field. The resolution of the sensors is approximately 0.1 nT. A disadvantage of the fluxgate sensors is that the magnitude of only one component of the total field can be measured, resulting in the output of smaller relative changes. Because of the direction of the earth’s magnetic field, this effect increases towards the equator. 4.3.2.1 Geoscan FM36 fluxgate gradiometer The Geoscan FM36 fluxgate gradiometer and earlier models of the same magnetometer are widely used in and purpose built for archaeological prospection.

36

Figure 13 A Geoscan fluxgate gradiometer with 0.5 meter sensor separation (left) and a Bartington fluxgate gradiometer with 1 meter sensor separation (right). The two fluxgate sensors are placed 0.5 meter apart in a lightweight aluminum or plastic case (Fig. 13). The magnetometer is carried with the sensors in a vertical alignment, and is sensitive to the vertical component of the earth’s magnetic field. The sensitivity of the instrument is 0.1 nT. The data quality strongly depends on the quality of the instrument setup, a procedure that is partly manual, partly electronic. The data that is collected is automatically displayed and recorded in the instrument’s memory. For this study, sites on which a Geoscan instrument was used were divided into 20 x 20 meter grid squares. Data was usually collected every 0.25 meter on lines with a 1 meter line sepa-ration, unless stated differently in the fact sheets (Appendix I). An automated trigger was used and grids were walked in a zigzag fashion. A setup and zeroing was preformed each time before commen-cing a new grid, in order to optimize the data quality. The data was downloaded each day to a laptop computer with the Geoplot program. 4.3.2.2 Bartington GRAD601 fluxgate gradiometer The Bartington GRAD601 fluxgate gradiometer is a new instrument that is becoming increasingly popular in archaeological prospection (Fig. 13). The sensor separation is 1 meter, which in theory makes the instrument more sensitive to magnetic variations at greater depths (Bartington & Chapman 2004). With this gradiometer, the vertical component of the terrestrial field is measured at a sensitivity of 0.1 nT. The setup is fully electronic, and is only preformed twice a day because the sensors are not very prone to drift. There is an option of attaching two gradiometers to the carrying harness, which increases the survey speed considerably. For this study the single gradiometer was used. The Bartington fluxgate gradiometer has become available during the course of this study. The proto-type of the instrument was tested next to the Geoscan instrument in Broekpolder and Harnaschpolder so that the results of the Geoscan and the Bartington instrument could be compared. Because the two datasets were fairly similar, they have not been displayed in this thesis. In addition, the Bartington instrument was used in this study on the site of Raalte by DW consulting (Appendix I, 19). 4.3.3 Magnetic susceptibility All magnetic susceptibility values in this study are mass related and are expressed as m3/kg. 4.3.3.1 Agico KLY-2 magnetic susceptibility bridge Soil samples for magnetic susceptibility measurements were taken by hand-auger (Boxmeer, Harnaschpolder, Heeten, Poeldijk, Smokkelhoek, Spalding, Stede Broec, and Uitgeest), directly from archaeological excavations (Beugen, Deil, Den Dolder, Ginneken, Grensmaas, Harnaschpolder, Heeten, Limmen, Meersen, Raalte, Uitgeest, Wervershoof, Wijk bij Duurstede and Zaltbommel) or without intrusion (Swalmen and Valkenisse).

37

For the hand augering a seven cm Dutch (screw)auger was used for top meter and a three cm gouge auger for samples deeper than a meter. Soil profiles were described and the remainder of the cores discarded after sampling. Archaeological features in excavation were sampled by pushing sample tubes into the exposed section or surface. These small plastic lidded tubes were also used for storing and transporting the samples. The soil samples were not dried in order to prevent any chemical changes by oxidation unless stated otherwise in Appendix I, but frozen and measured within 14 days from sampling, using an Agico KLY-2 Kappabridge in the palaeomagnetic laboratory of the Universiteit van Utrecht. 4.3.3.2 Bartington MS2B susceptibility meter On the sites of Broekpolder and Harnaschpolder, the first set of samples for magnetic susceptibility analysis were taken with a three cm gouge auger. The first five cm of topsoil were always discarded. Samples were taken from the auger with a spatula and stored in polyester ziplock bags. Samples were air dried on paper plates, and ground with a porcelain mortar and pestle. The magnetic susceptibility of the samples was measured on a Bartington MS2B AC magnetic susceptibility bridge in the Department of Archaeological Sciences in the University of Bradford, UK. Measurements were repeated three times and the mean measurement was calculated. The AC sus-ceptibility bridge was calibrated using a sample of high alumina cement and a sample of manganese sulphate. The samples were weighed on a top-pan balance. This was repeated twice and the mean weight of the sample was calculated. 4.3.3.3 Fractional conversion A heat treatment was given to a selection of soil samples from Broekpolder and Harnaschpolder in order to obtain a value for maximum conversion. The procedure as described by Clark (1996) was followed. Approximately 10 ml of each sample was weighed on a top-pan balance. The volume of the samples was measured using a 25 ml cylinder. The magnetic susceptibility of the samples was measured on a Bartington MS2B AC magnetic susceptibility bridge in the Department of Archaeological Sciences in the University of Bradford, UK. The measurement was repeated and the mean magnetic susceptibility value was calculated. One gram of plain flour was added to each sample and the samples were placed into porcelain crucibles and covered with a porcelain lid. A Carbolite electric muffle kiln was heated up to 650 ºC with the chimney closed. Once this tempe-rature was reached the samples were placed in the furnace and left in the furnace for an hour after the kiln had reached a temperature 650 ºC again. The furnace was switched off and left to cool for half an hour. After cooling the samples were taken out. The lids were removed and the samples were stirred with a wooden spatula. The furnace was switched on again with the chimney open. After reaching 650 ºC the samples were put back into the furnace and heated for 45 minutes after the furnace reached the temperature of 650 ºC again. The furnace was then switched off and left to cool for half an hour before taking the samples out. After reaching room temperature, the volume of the samples was measured using a 25 ml cylinder. The samples were weighed using a top-pan balance. The magnetic susceptibility of the samples was measured on a Bartington MS2B AC magnetic susceptibility bridge. The measurement was carried out in triplicate and the mean magnetic susceptibility was calculated. 4.3.4 Thermomagnetic, IRM and ARM measurements 4.3.4.1 Thermomagnetic measurements Soil samples were obtained either by hand-augering (Harnaschpolder and Spalding) or from the sections or surfaces of archeological excavations (Harnaschpolder south, Limmen), one set of samples came from a freshly cut ditch (Broekpolder). Samples were stored in lidded plastic sample tubes, and were either freeze dried (Harnaschpolder and Spalding) or air dried (Broekpolder, Harnaschpolder south and Limmen). The Harnaschpolder and Spalding samples were taken from the reducing part of the soil section and were freeze dried in order to prevent any chemical changes in the soil samples that could possibly influence the iron mineralogy. A set of eight samples was measured for magnetic sus-ceptibility before and after freeze drying (Table 7).

38

A slight change (maximum value 4.4%) can be observed. It is likely that part of the material has oxidized during the freeze drying process, probably the material on the surface of the sample that was exposed to air. It is assumed that the remaining material is chemically unchanged. The air dried samples were taken from the oxidizing part of the soil section and it was thought that further air drying would not cause any chemical changes in the soil samples. Prior to the thermomagnetic runs, the magnetic susceptibility of all samples was measured using a AGICO KLY-2 susceptibility bridge in the Palaeomagnetic Laboratory ‘Fort Hoofddijk’ of the Universiteit van Utrecht. Table 7 Magnetic susceptibility before and after freeze drying of a selection of soil samples from Spalding.

depth magnetic susceptibility x 10-8 m3/kg wet

magnetic susceptibility x 10-8 m3/kg after freeze drying

% change

A6 80 68 69 1.47 A6 300 1138 1112 -2.28 A6 320 390 375 -3.85 A6 330 382 377 -1.31 A6 410 250 248 -0.80 A7 200 90 94 4.44 A7 350 245 241 -1.63 A7 500 79 82 3.80 Thermomagnetic measurements were carried out with a modified horizontal-translation-type Curie Balance (Mullender et al. 1993) with a sensitivity of approximately 5 x 10-9 Am2 in the Palaeomag-netic Laboratory ‘Fort Hoofddijk’ of the Universiteit van Utrecht. Samples were placed in a quartz glass sample holder and kept in place with quartz glass wool, and heated and cooled in air in 16 runs of 15-150, 150-50, 50-250, 250-150, 150-300, 300-200, 200-350, 350-250, 250-400, 400-300, 300-500, 500-400, 400-600, 600-500, 500-650, 650-15 °C, at a heating rate of 10 °C and a cooling rate of 15 °C per minute. The alternating field varied from 150-300 mT. 4.3.4.2 Anhysteretic remanent magnetisation (ARM) measurements Samples from Broekpolder and Harnaschpolder were selected for ARM and IRM measurements. Approximately 0.2 gram of material was placed in a standard sized sample holder with approximately 0.2 gram of CaCO3 to prevent sticking. The sample was homogenised with epoxy resin. The samples were left to dry overnight. A measurement in four directions was carried out in an Agico JR-5A spinner magnetometer. After this initial measurement the samples were demagnetised in the AF demagnetiser. An anhysteretic remanent magnetisation was impaired in the samples with a steady field of 1640 mT, and a peak alternating field in steps of increasing strength: 0, 2.15, 4.05, 6, 8, 10.1, 15, 20, 25, 30.05, 40.05, 50, 66, 80, 100, 130, 151, 176, 200, 250.05 and 260 mT. The samples were measured in four directions with the spinner magnetometer after every step. 4.3.4.3 Isothermal remanent magnetisation (IRM) measurements The samples that were used for the ARM measurement were again demagnetised using the AF demagnetiser. Using a PM4 pulse magnetizer IRM was acquired in the samples in steps of increasing steady field strength: 0, 10, 15, 20, 25, 30, 40, 50, 65, 80, 100, 120, 150, 180, 200, 250, 300, 400, 500, 650, 800, 1000, 1200, 1400 and 1600 mT. The samples were measured with a JR-5A spinner magnetometer after every step. 4.4 Software Two geophysical software packages were used for downloading the geophysical data from the instrument and for processing data, Geoplot and Archeosurveyor, the latter has become available in the course of this study. 4.4.1 Geoplot The Geoplot software package was built for use with the Geoscan instruments, like the FM36 that was used in this project.

39

Data is downloaded directly from the instrument in grids that generally measure 20 x 20 meter. The grids are compiled in a so called master grid after which a composite dataset is obtained. The data that was collected for this study has undergone minimal processing. Generally, datasets have been despiked and edge-matched. Sometimes a de-stagger routine had to be applied to data that was collected in a zig-zag manner. In certain case studies inter- or extrapolation of the data was necessary for the clarity of the output picture, as directional differences in data density can give the data a ‘messy’ appearance. Datasets that had been collected on a 1 x 0.25 meter or a 1 x 0.5 meter spatial resolution were generally inter- and extrapolated to 0.5 x 0.5 meter. For every dataset is has been assessed if the application of filters was beneficial for the quality of the output picture. In a few instances this was the case, if a filter has been applied, mention is made in the caption of the figure. A greyscale palette with 55 shades of grey has been used to display the data throughout this study. The graphics are exported from the program as bitmap files. The majority of the sites that were surveyed in this study were processed with Geoplot, except for Swalmen, Heeten (2004 survey), Meteren and Ossenisse (2005 survey) for which Archeosurveyor was used. 4.4.2 Archeosurveyor The Archeosurveyor software imports directly from the instrument in grids. These grids are as-sembled in a grid assembly, the resulting dataset is a composite. Processing routines like despike, edge match, destagger and interpolation have been used in a similar fashion as in the Geoplot program. Filters have not been used. A greyscale palette with 99 shades of grey has been used in Archeosurveyor to display the data. Graphic files have been exported as portable network graphics (png) files from Archeosurveyor. 4.4.3 Further software Graphics from the geophysical software have been georeferenced to the Dutch national grid (Rijks-driehoeksnet) and imported into Autocad, where the geophysical data was linked with other sources of information like archaeological data and topographical maps. The Autocad drawing has been exported as a graphic pdf file. The final lay out of the figures in this thesis has been made in Adobe Illustrator. The Excel spreadsheet program has been used to calculate and analyze the magnetic susceptibility data. For the analysis of the Curie Balance measurements, the Cursmooth 2.03 program which was developed by the Palaeomagnetic Laboratory of the Universiteit van Utrecht was used (Mullender et al. 1993). From the same laboratory is the program IRM-CLG 1.0 (Kruiver et al. 2001, Heslop et al. 2002) that was used to analyze the IRM data that was obtained.

40

41

5 Estuarine deposits 5.1 Introduction This chapter investigates the possibilities for magnetic prospection on estuarine soils. The first section is concerned with the detection of a specific group of archaeological features; peat extraction pits and associated peat ties. Archaeological prospection by means of hand augering can be hampered by past peat extraction activities, and the use of magnetometry in order to map disturbances in the peat layer is assessed. The second section focuses on magnetic prospection on estuarine soils in general and on the occurrence of large, strongly magnetic anomalies, which probably have a geological origin. In the third section detailed mineral magnetic investigations are used in order to understand the iron mineralogy of archaeological deposits on two archaeological sites on estuarine geology. Moreover, the cause of the highly magnetic possibly geological anomalies is investigated here. 5.2 The mapping of a peat-extraction landscape1 5.2.1 Introduction The use of peat for fuel or possibly for the production of salt in the past is a well known practice in the western part of the Netherlands, and indeed in other countries where peat layers occur close to the surface2. The process of peat excavation has had an enormous impact on the landscape. Most obviously, land that was exploited for peat is now lower lying and wetter than adjoining unexcavated areas. Because this subsidence and water logging of the land caused a problem for the stability of the dykes, digging for peat was forbidden in some areas in the Netherlands in the Middle Ages and in later periods (Van Vliet 2007). For more than one reason scientists from different fields are interested in the geographical component of peat excavation. This paper will start with setting out these reasons. Then the techniques that are generally used to map past landscapes and their possibilities and pitfalls in the framework of a peat exploitation landscape are discussed. In the core of this paper two geophysical techniques are introduced that are less commonly used in the investigation of the historical landscape in the Netherlands, and two case studies are presented. In the conclusions a combination of conventional and novel techniques is proposed for the detailed investigation of this past landscape. 5.2.2 The problem The process of turfing has left its traces in the landscape, generally the land has subsided and become wetter, but on a smaller scale a number of features that relate to the peat extraction can be recognized. These can be divided into three categories: (i) peat ties; pits and ditches that were dug to excavate the peat from (ii) barrow ways; rims of unexcavated peat that were left in place and were used as barrow ways or to

dry the peat on (iii) water ways; canals and ditches that were dug to transport the peat.

1 This paragraph has been adapted from Kattenberg (2007) with minor textual alterations. 2 e.g. in the United Kingdom (Somerset, Wales), Ireland, and Germany. In the Roman period, Plinius already

writes about the use of peat as a fuel of the Chauci in northwestern Germany, the account reads: they fashion the mud, too, with their hands, and drying it by the help of the winds more than of the sun, cook their food by its aid, and so warm their entrails, frozen as they are by the northern blasts (Gaius Plinius Secundus, Naturalis Historia XVI 1.1, translation John Bostock, London, Taylor & Francis 1855).

42

Both historical geographers and archaeologists are interested in the lay-out of these features. Both groups are of course interested in the human impact on the landscape, and mapping the traces of past peat extraction can help to understand the process of turfing. But unfortunately the features that can be associated with peat extraction can rarely be seen in the landscape; most peat exploitation landscapes are covered with sediments from post-exploitation marine inundations. The sediments have filled the negative features and generally flattened the relief. There is another reason why archaeologists require detailed geographical information about past turfing. In archaeology, information about past cultures is usually found in strata, occupational layers in the ground. These layers can have been damaged or destroyed in the process of peat excavation (Fig. 14).

Figure 14 Archaeological remains can be damaged by peat extraction. A: archaeological remains of habitation in the upper levels of the peat layer are covered with clastic sediments; B: peat extraction has disturbed archaeo-logical levels, but archaeological information is still preserved in the barrow ways; 1: clastic sediment; 2: peat. An interesting situation occurs because an activity that is archaeologically interesting in itself, has at the same time destroyed archaeological remains of earlier periods. All that is left of the archaeological archive in a landscape that has been exploited for peat is contained in the unexcavated parts of the landscape, and in most cases this is equivalent to the barrow ways. The excavation in Ellewoutsdijk by the ADC (Sier 2003) showed how destructive turfing can be for the archaeological physical archive, but also how much information can still be extracted from the unexploited parts of the landscape. Here the remains of a Roman settlement that was heavily damaged by Medieval peat extraction activities were excavated, but the damaged dataset still allowed for a comprehensive analysis of the settlement, not in the least because of the excellent conservation of organic materials in the peat. And here it seemed that the Medieval excavation was in a way influenced by the earlier occupation, as the posts that were used in the Roman buildings were generally left in place and barrow ways were made in their surroundings (Fig. 15). The excavation of the large timbers probably cost too much time and effort for the Medieval peat diggers. Most peat extraction activity was ad hoc and small scale, individuals or small groups cut peat as it was needed, and an over-all organization of the location, dimensions and orientation of the peat ties was often not present. This has resulted in a very irregular and patchy appearance of the traces of peat exploitation in many places (Fig. 16), and for this reason the lay-out of the features is difficult to predict. If non-destructive archaeological prospection methods can be used to investigate the landscape, elements from the historical geographical landscape can be mapped, and, from an archaeological point of view, the potential for the presence of archaeological remains can be assessed.

43

Figure 15 A row of posts from a Roman building is left in place by the Medieval peat excavators. Photo from archaeological excavation Ellewoutsdijk (ADC Archeoprojecten). From Sier 2003, with permission.

Figure 16 An example of the irregularity of the peat ties from which the peat was extracted. Most peat extrac-tion was ad hoc and small scale. An example from Gwent, Wales. (http://www.ggat.org.uk/excav_A559_gwent%20euro%20(wilkinsons).htm on 14/11/2005). This section investigates new methods to map the past landscape in a non-destructive way. As far as the author is aware, there are no other examples of the use of geophysical methods to map peat extraction in The Netherlands or abroad. 5.2.3 Methods for the investigation of former peat extraction Historical maps and documents can give information about the location of areas where peat extraction has taken place in the past, and in some cases waterways and roads that can be associated with the exploitation of the peat landscape can be recognized. Individual peat ties or barrow ways, however, will generally not have been mapped in the past, and it is these features that both archaeologists and historical geographers are interested in for reasons outlined above.

44

Traditionally, there are three methods that can be used for the investigation of the features that are associated with peat extraction; excavation, remote sensing and hand augering. A full archaeological excavation of a landscape that has been disturbed by past peat extraction will yield the most detailed information about the way that the landscape and the resources within it have been used in the past. Geographically, excavation will give information about the size, location and orientation of the individual peat ties and the water and barrow ways in between. A high level of detail of the infor-mation that is collected can be achieved this way, but more often than not it is not feasible to fully excavate traces of peat extraction, because archaeological excavations are very time consuming. Remote sensing techniques are considerably faster in their use, and include aerial photography, satellite imagery and digital elevation information that has been collected from the air. Traces of peat extraction features can sometimes be seen on conventional aerial photographs, and if they are, they can be as easy to resolve as from excavation. This technique, however, is constrained by the presence of sediments of post peat extraction marine inundations that have filled in the peat ties, have covered the features and have evened out the relief that could be indicative of the presence of peat extraction features at depth. In some cases the presence of features like peat ties and barrow ways can be proofed, but their absence cannot. Other remote sensing techniques can work theoretically, but have not been used in the research into the peat exploitation. Infrared imaging techniques rely on the difference in heat retaining properties in different materials. It can be expected that peat (barrow ways) and silt (fill of peat ties) indeed do differ, but like conventional aerial photography, infrared imaging is only concerned with very superficial features, and any layers covering the original surface will hamper the detection of subsurface features with this technique. The same is true for the use of airborne elevation measurements, theoretically the underground presence of peat ties, and in contrast unexcavated rims of peat, can be expected to influence the superficial micro relief. In practice layers of marine sediments can hide much or all of the relief that is present in the subsoil, and again this technique can under certain circumstances be used to indicate the presence of features, but never the absence. Hand augering is a method that is commonly used in archaeological prospection in the Netherlands. Whether or not peat extraction activities are recognized depends mainly on the size and type of grid that is used for the coring. Peat ties can easily be mapped because of the disturbed nature of their fill and because the top or all of the layer of peat has been taken away. Barrow ways on the other hand cannot be recognized as such, because here the soil profile is undisturbed. They can only be recognized in contrast to the peat ties. It is difficult to find the right grid size for the investigation of a peat extraction landscape. On the one hand the scale of the exploitation is often large, whereas the individual features are of a small scale and have an irregular nature. In general, a tight grid like 20 x 20 meter will pin point the areas where peat has been extracted, but this will not allow individual features to be mapped. 5.2.4 The use of geophysical methods for the investigation of former peat extraction Geophysical methods are a group of techniques that map variations in the physical properties of the subsoil. They have been developed in earth sciences, but have since the 1950s been adapted for archaeological prospection. The most commonly used methods are magnetometer surveys, electrical resistance surveys, and ground penetrating radar. For details about these methods see Gaffney and Gater (2003). Magnetic methods Magnetic methods that are used in archaeological prospection map magnetic contrasts in the field. As every material has different material properties, two materials adjoining each other in the horizontal plain in principle can be mapped magnetically. Archaeological features like pits and ditches generally have a fill that consists of a different material than the matrix that they have been dug into. The magnetic contrast that this is causing can generally be measured on the surface, and the shape and size of the archaeological feature can be assessed without excavation. Thinking along these lines, there is a material difference between the fill of the peat ties with mainly clastic materials on the one hand, and the barrow ways and unexcavated peat on the other.

45

The theoretical magnetic contrast that exists between peat and clastic material was found to be a real and considerable contrast on the site of Harnaschpolder, Zuid-Holland (Appendix III). On this site, peat turned out to have a magnetic susceptibility close to zero, whereas the clastic materials had a magnetic susceptibility that were generally higher. It can be assumed that in similar, estuarine, geolo-gical circumstances, similar magnetic contrasts do exist. In order to test the hypothesis that elements in buried peat exploitation landscapes like peat ties and barrow ways have a magnetic contrast that can be measured at the surface, magnetometer surveys were carried out on the site of Smokkelhoek, where the presence of past peat exploitation activity was known by hand augering (Jansen 2003). Electrical methods The way electrical methods can be used in archaeological prospection is also based upon contrasts, in this case the methods rely on contrasts in resistivity or conductivity of the soil. These contrasts are almost entirely dependent on the moisture retaining properties of the soil, and in relation to this, on the grain size and packing of the material. As is the case with magnetic methods, electrical methods can map the (horizontal or vertical) interface of two materials with different resistivities. Peat can almost be entirely waterlogged, whereas in clastic materials only the voids in between the grains that make up the soil that can retain water. This causes clastic materials to have a higher electrical resistivity than peat. Hence we assumed that elements in buried peat exploitation landscapes like peat ties and barrow ways have an electrical contrast that can be measured at the surface. In order to investigate this hypothesis an earth resistance survey was carried out in Kolhorn, on a location that was known to have had past peat exploitation activity. 5.2.5 Sites Geology The Holocene development of the Dutch coastal plains (Fig. 17) is characterised by an alternation of clastic deposition under marine or supratidal environments, and peat formation. In general, the sequence consists of a layer of basal peat overlying Weichselian coversands, followed by two major phases of clastic sediments, respectively termed Wormer and Walcheren3 deposits. These deposits are separated by the so-called Holland peat. It is this peat layer that was excavated for fuel and possibly for salt production in the Middle Ages.

Figure 17 The extent of the marine and estuarine depo-sits in the Netherlands. S = Smokkelhoek, K = Kolhorn.

3 Recently a new lithostratigraphic nomenclature has been introduced (De Mulder et al. 2003). Wormer corres-

ponds with Calais in the old terminology, Walcheren with Dunkirk. They are part of the formation of Naald-wijk.

46

Smokkelhoek The archaeological site at Smokkelhoek (Fig. 17) was first inhabited in the Roman Period. At this time the Holland peat was still being formed. Marine influence increased from the late Roman Period, and transgressions of the Walcheren phase continued to take place until as late as the 1950s. These transgressions were more severe in this area than in the Northern part of the country, and a layer of sediment with a thickness of a few meters in places was deposited on top of the peat, eroding it in places. In the Middle Ages large-scale peat extraction became an important industry. Pits were dug through the layer of Walcheren sediment in order to reach the peat. By digging into the peat, traces of Roman occupation were erased as well. After the peat extraction, the pits were not back-filled, but left open to be filled in with the sediments of later marine transgressions. Kolhorn Between the 8th and the 14th century the peat area around the archaeological site in De Waardpolder North of Kolhorn (Fig. 17) was inundated with sea water. After these inundations large scale peat extraction has taken place in the area, the traces of which are still visible in the landscape today (Fig. 18), on the ground but also on aerial photographs. One of the features that is associated with the peat extraction in this area is the ‘road of Paludanus’, an almost 5 km long linear feature that was first recognized by Rutger Paludanus in 1776. He interpreted it as being a road on a dyke. The feature is still visible from the air, but due to ploughing, less so on the ground. Archaeological excavations in 1995 have shown that the road is in fact the last strip of unexcavated peat in the area. It is quite likely that it was deliberately left unexcavated so it could be used as a dyke. Van Geel and Borger (2002) show that the peat extraction is probably associated with industrial salt production. The name of a nearby farm ‘goud na zout’ (‘gold after salt’) could support this interpretation. 5.2.6 Methodology Smokkelhoek On the site of Smokkelhoek an area of 40 x 40 meter was selected for the magnetometer survey. This selection was based upon a larger scale magnetometer survey with a resolution of 1 x 0.25 meter (Kattenberg, in prep.). The spatial resolution of the survey was 0.25 x 0.5 meter, the instrument that was used was a Geoscan FM36 fluxgate gradiometer with an instrument resolution of 0.1 nT. In order to support the interpretation of the magnetometer survey, hand augering was carried out with a 7 cm Dutch (screw) auger and a 3 cm gauge auger.

Figure 18 Traces of peat extraction in Polder Wieringerwaard, to the west of the Waardpolder and 1 km west of the village of Kolhorn.

47

Kolhorn In Kolhorn an electrical resistance survey was carried out with a TRCIA resistance meter with a spatial resolution of 1 x 1 meter and an instrument resolution of 0.1 Ohm. The location of the field-work was chosen with an aerial photograph that clearly showed the road of Paludanus as a soil mark (Robas 1986). The area that was investigated measures 20 x 80 meters and was planned to be perpendicular to the road. The probe distance between the mobile probes was 1 meter, and the depth of investigation can be assumed to be approximately 1 meter. In order to aid the interpretation of the geophysical data, hand augering was carried out with a 7 cm Dutch (screw) auger and a 3 cm gauge auger.

Figure 19 The results of the magnetometer survey on the site of Smokkelhoek. 5.2.7 Results Smokkelhoek The results of the magnetometer survey in Smokkelhoek are displayed in Figure 19. The results can be thought of as a detailed map of the magnetic contrasts in the soil. Negative anomalies are displayed in black, and positive anomalies in white. Based on laboratory measurements it was expected that barrow ways and other remnants of peat would cause a negative magnetic anomaly. For this reason the lighter coloured structures are interpreted as peat ties, the darker linear structures as rims of peat. A schematic representation of the data is given in Figure 20. A number of cores was carried out along the line C-C’ to check the interpretation of the magnetometer data. It was found that the location of the negative magnetic anomaly that was investigated by hand augering did correspond with the presence peat in the upper part of the soil section, whereas in cores on the location of the positive anomalies only a thin layer of peat was found, starting at greater depth. Because of the way peat grows, the top of a peat layer is generally horizontal. It is likely that the ‘depressions’ in the peat layer as mapped during the coring is the effect of peat extraction. Kolhorn4 The result of the earth resistance survey in Kolhorn is displayed in Figure 21. Zones of higher resis-tance have dark colours, zones of lower resistance lighter colours. Two areas with high resistance values appear to be anomalous to the rest of the area, a linear anomaly on the east side, and a zone of high values in the middle of the surveyed area. The latter group of anomalies has a southwest-northeast direction and consists of strips of higher and lower resistance.

4 The author would like to thank the Melchior family for allowing us to survey the Kolhorn site.

48

Figure 20 Schematic representation and interpretation of the results of the magnetometer survey on the site of Smokkelhoek. Overall, the width of the zone of high resistance measures approximately 15 meters. The location, width and direction of the anomalies correspond to the soilmark that was visible on the aerial photo-graph and represents the road of Paludanus. The eastern anomaly is related to a ditch that was partly filled in but could still be seen as a depression on the surface. A number of cores was carried out in order to map the area. A typical profile consists from top to bottom of a patchy dark layer of approximately 40 cm of mixed materials (silt, clay and sand) with occasional (layers of) shell, followed by a 20 cm layer of peat. Under the peat is a sequence of silty, sandy and clayey light grey estuarine deposits, the reduction zone starts at 150 cm. There are no boreholes positioned over the high resistance anomalies.

Figure 21 The results of the earth resistance survey on the site of Kolhorn. 5.2.8 Discussion The material contrast between the unexcavated peat and the clastic fill of the peat ties appears to be sufficient to be mapped with geophysical methods. This is the case for the two surveys that were presented in this chapter, and is likely to be true for other peat extraction landscapes in similar settings. In the Dutch coastal plain peat ties are generally filled with clastic material, creating a mate-rial contrast that can be successfully mapped. Peat often contains a mineral fraction, and the increase in volume of this mineral component causes the two materials to become more alike, which will lead to a decrease in material contrast between the peat and the clastic material.

49

The magnetic contrast between the two materials was previously investigated in the laboratory, where it was found that peat has a magnetic susceptibility close to zero, whereas clastic material has a slightly positive magnetic susceptibility. This difference in magnetic susceptibility is reflected in the results of the magnetometer survey in Smokkelhoek, where remnants of peat show as negative anomalies. The magnetic contrast is not very large, but the volume of the features is quite substantial, which aids their detection. The peat ties could barely be identified in the coarsely spaced survey (Appendix I, 21) of 1 x 0.25 meter, but are well defined in the 0.5 x 0.25 meter survey. This latter grid spacing is recommended for the investi-gation of peat extraction landscapes. In the electrical resistance survey, the remnants of peat that were still present in the subsoil could be mapped because they had a higher resistance than the matrix that they were embedded in. This contrast was reverse to the hypothesis. Because peat has a large capacity to retain water, it was expected that the inclusion of peat in the matrix causes a lower soil resistance than would have been obtained in a purely mineral soil. The feature that was mapped is of larger scale than the barrow ways that were investigated in Smokkelhoek, and it may have to be classified more as a road than as a barrow way. The high electrical resistance may be caused by a certain metalling of the road or by the compaction of the top layer of the feature during its use. The most likely explanation, however, is that the peat matrix has dried out, something that can be supported by the hand augering data, the peat was found in the oxidizing part of the soil section. The air in the pockets where the soil water was first stored hampers the electrical current flow and causes the material to have an overall high electrical resistivity. The contrast between peat and clastic material can probably be mapped when the peat is waterlogged or dried out. In a semi-wet state peat may have similar properties to a completely mineral soil. The 1 x 1 meter resolution that was used in the Kolhorn survey was sufficient to map the road of Paludanus and details within it. This resolution is generally high enough to map linear features with a width of up to 0.5 meters. For narrower features a higher resolution is required, e.g. 1 x 0.5 meter. 5.2.9 Conclusion In Smokkelhoek and Kolhorn electrical and magnetic methods proofed to be able to map features that are associated with peat extraction. In general contrasts between peat and clastic materials like sand, silt and clay can be detected with magnetic and with electrical methods. Exceptions occur when the peat has a high proportion of clastic material, and when the peat is semi-wet. Geophysical surveys are non-destructive to the archaeological record and can be carried out much faster than archaeological excavations. The information obtained is very detailed and for information about the shape in plan of individual features, the level of detail approaches that of excavation. This detail is also the advantage of geophysical methods over prospection by hand augering, whereas the latter method gives a type of vertical and ‘ground truth’ information that cannot be obtained with geophysical methods. The application of different prospection methods on one site will always lead to a better under-standing of the buried features, but is not always feasible. The choice of method(s) needs to be based on the merits of all methods individually, and type of information obtained (e.g. horizontal or vertical), resolution and speed of investigation do have to be considered. In historical landscapes in general, hand augering is a prospection method that can identify surfaces and layers that are associated with previous use and habitation of the landscape, but generally not individual features. In a peat exploitation landscape the general area in which peat extraction has taken place can be identified, but not the individual features like peat ties and barrow ways. This level of detail can be obtained in plan with geophysical methods. For a 3D investigation of these features however, excavations remain necessary. Geophysical methods can be used to bridge the gap between coring and excavation, both in the detail of the information obtained and the in speed of investigation.

50

5.3 Archaeological prospection of the Dutch estuarine landscape by means of magnetic methods5

5.3.1 Introduction Magnetic variations in hard rock geology are known to hamper magnetometer surveys carried out for archaeological purposes (e.g. Clark 1996), but less attention has been paid to the magnetic response of geological structures in unconsolidated sediments (for an example see Weston 2001). If geological magnetic variations are as large as or larger than archaeologically caused variations, the latter may not be distinguishable from the natural. Smaller scale geological features thus cause a bigger problem than broad geological variations. Long wavelength or gradual geological changes can be filtered out by post-processing software, or by using the magnetometer in a gradiometer configuration. Short wavelength, abrupt geological changes, however, can have magnetic characteristics very similar to buried archaeological features, which makes it difficult to separate the two. The geological subsoil of the western part of the Netherlands consists entirely of unconsolidated Quaternary sediments. As the coastal plains occupy a considerable part of the Dutch landscape (Fig. 22) and parts of this landscape have been inhabited continuously since the Neolithic, it is important to understand the magnetic characteristics of the marine and the estuarine (salt marsh) deposits. In the current study several known sites dating from various archaeological periods have been investigated. This paper reports on preliminary findings. The results of magnetometer surveys on two of these sites, Harnaschpolder and Smokkelhoek (label-led H and S respectively in Fig. 22) present several problems that may be typical of this specific geological setting. Given the widespread occurrence and the significance of estuarine sediments as a geological substrate in The Netherlands, it is necessary to investigate these problems. The aim of this paragraph therefore is to: (i) identify the possible reasons for the lack of magnetic contrast that was expected between the fill

of archaeological features and the undisturbed matrix on the sites discussed; (ii) discuss the (geological) origin of the magnetic anomalies encountered in Harnaschpolder and

Smokkelhoek. The paragraph starts with an overview of the geological development of the estuarine landscape of The Netherlands, and the archaeology of the sites discussed. This is followed by the methodology and the results of the magnetometer survey and the magnetic susceptibility sampling in Harnaschpolder and Smokkelhoek.

Figure 22 The extent of marine and estuarine deposits in The Netherlands. H = Harnaschpolder, S = Smokkel-hoek.

5 This paragraph has been adaped from Kattenberg and Aalbersberg (2004) with minor textual alterations.

51

5.3.2 Archaeology and geology The Holocene development of the Dutch coastal plains is characterised by an alternation of clastic deposition under marine or supratidal environments, and peat formation. In general, the sequence consists of a layer of basal peat overlying Weichselian coversands, followed by two major phases of clastic sediments, respectively termed Calais and Dunkirk6 deposits. These deposits are separated by the so-called Holland peat. The area of Harnaschpolder (Fig. 22) has been occupied since the Neolithic period. Although the Harnaschpolder was inhabited almost continuously, the landscape changed dramatically between the different periods, resulting in a layered archaeological landscape. In the Neolithic period, settlements were located on dunes along the coast; these were high points in a landscape dominated by wetlands. After the Neolithic, marine influence increased, and the landscape was covered with silts and sands (Calais phase). The tidal flats of this period gradually developed into tidal marshes. As the marine influence diminished, vegetation eventually returned, resulting in the growth of a layer of fen peat (Holland peat). At the start of the Iron Age, the peat blanket had locally reached a thickness of several meters. The next phase of marine transgressions occurred from the Iron Age to Late Medieval times. These Dunkirk transgressions partly eroded the peat before covering it with another layer of silts and sands. In this new tidal marsh, Iron Age and Roman Period people would inhabit and farm the raised sandy beds of former tidal creeks. Medieval occupation consisted of dwelling mounds constructed in the areas where the peat had not been eroded away. A last phase of marine transgressions, in the Late Middle Ages, covered the already layered archaeological landscape with another thin layer of marine sediment. At Smokkelhoek (Fig. 22) the site was first inhabited in the Roman Period. At this time the Holland peat was still being formed. Marine influence increased from the late Roman Period, and trans-gressions of the Dunkirk phase continued to take place until as late as the 1950s. These transgressions were more severe in this area than in the Harnaschpolder, and a layer of sediment with a thickness of a few meters in places was deposited on top of the peat, again eroding it in places. In the Middle Ages people inhabited the drier parts of the landscape, for example, the former tidal creeks, just like in the Roman Period in Harnaschpolder. Large-scale peat extraction became an important industry. Pits were dug through the layer of Dunkirk sediment in order to reach the peat that was either used as a fuel, or as a source of salt. By digging into the peat, traces of Roman occupation were erased as well. After the peat extraction, the pits were not backfilled, but left open to be filled in with the sediments of later marine transgressions. All through the archaeological periods discussed here, houses would have been timber built; stone masonry was not used because it simply was not available, and brick was only introduced in the late Middle Ages. The archaeological record thus mainly consists of in-filled pits and ditches, and of post-holes. The fill of the archaeological features generally has a different colour and a different texture compared to the surrounding matrix. On both sites the Roman and Medieval features are contained within the fist meter of the matrix. Lacking bedrock and gravels, any magnetic anomaly caused by archaeological features would have to be either the result of differences in soil magnetic susceptibility, or, for features like hearths and kilns, would have to be of a thermoremanent nature. On the other hand, magnetic anomalies that are caused by changes in geology may be due to differ-rences in magnetic susceptibility, or differences in natural remanent magnetization (NRM). 5.3.3 Methodology Magnetometer surveys were carried out at both Harnaschpolder and Smokkelhoek. A Geoscan FM36 Fluxgate Gradiometer was used at a spatial resolution of 0.5 x 1.0 meter, and an instrument resolution of 0.1 nT. Soil samples for magnetic susceptibility measurements were taken by hand-auger. For the top meter a 7 cm Dutch (screw)auger was used and a 3 cm gouge auger for samples deeper than a meter. Soil profiles were described and the remainder of the cores discarded after sampling. Archaeological features in excavation were sampled by pushing sample tubes into the exposed section.

6 Recently a new lithostratigraphic nomenclature has been introduced (De Mulder et al. 2003). The now obso-

lete Calais-Dunkirk terminology is still being used frequently, and to aid reference to older literature, this paper will refer to the ‘old’ terminology.

52

These small plastic lidded tubes were also used for storing and transporting the samples. The soil samples were not dried in order to prevent any chemical changes by oxidation, but frozen and measured within 14 days from sampling, using an Agico KLY-2 Kappabridge in the palaeomagnetic laboratory of the Universiteit van Utrecht. The weakly diamagnetic contribution of the sample tubes (~ -10 x 10-8 m3/kg) was taken into account. A heat-treatment was carried out on a selection of samples from Harnaschpolder in order to obtain fractional conversion values. This set of samples was air-dried, and the samples were crushed with a porcelain mortar and pestle. The magnetic suscepti-bility of the samples before and after the treatment was measured on an AC magnetic susceptibility bridge in the Department of Archaeological Sciences in the University of Bradford, UK. One gram of plain flour was added to each sample and the samples were placed into porcelain crucibles and covered with a porcelain lid. A Carbolite electric muffle kiln was used and the procedure similar to the one described by Clark (1996) was followed. 5.3.4 Results The most striking magnetic anomalies that were encountered at Harnaschpolder (Fig. 23) are two bands of positive anomalies with a smaller negative component, more or less at right angles to each other. Between these two bands smaller linear positive anomalies can be seen. Test trenches that were excavated prior to the magnetometer survey overcut these magnetic ano-malies, but no apparent relation was found between the archaeological features that were excavated and the anomalies. In fact, none of the archaeological features that were encountered in the test trenches could be identified by the gradiometer survey. In de Smokkelhoek (Fig. 24) linear and curved positive anomalies occurred in pairs. In the north-western corner of the surveyed area features seem to be overcutting each other. The only unpaired anomaly, running north-south in the eastern part of the survey, could be identified as a track also visible on the 1832 map of the area. Patches of noise correlate to the field boundaries on the same map and are probably caused by the material in the in-filled ditches. In the centre of the eastern block of survey, faint linear negative anomalies indicate the presence of rims of peat in the subsoil. These are the bands that were left in during the peat extraction. In neither of the sites any magnetic response of archaeological features could be identified. Test trenching (in Harnaschpolder) and hand-augering (in Smokkelhoek) had however positively identified archaeological features in the surveyed areas. Soil samples were taken from two excavations directly to the North and the South of the surveyed area in Harnaschpolder. Samples were taken from pits, ditches and a posthole from the Roman Period, as well as from the undisturbed silts and the topsoil. Magnetic susceptibility measurements show that generally the values are very low (Table 8). There is very little difference between the magnetic susceptibility of the fill of the Roman Period features and the natural material they are embedded in. Values for the topsoil, on the other hand, are higher, and less consistent.

Figure 23 The results of the magnetometer survey at Harnaschpolder with the location of core A.

53

Figure 24 The results of the magnetometer survey at Smokkelhoek with the location of core B. In order to assess if the sediments are capable of acquiring higher magnetic susceptibilities, samples were taken at Harnaschpolder for an investigation into the fractional conversion. Sixteen samples were taken from a hand auger so that all sedimentary layers in the profile were represented. Fractional conversion values were calculated as χ/χmax x 100%, and range from 0.31% to 6.08% with an average of 3.36%. This indicates that the low magnetic susceptibility values encountered are not caused by a lack of iron; the magnetic susceptibility thus has to be related to the type of iron compounds present in the soil. Table 8 Magnetic susceptibility of soil samples from Harnaschpolder excavations. N: the number of samples; SD: the standard deviation from the mean magnetic susceptibility. N mean magnetic susceptibility

x 10-8 m3/kg SD x 10-8 m3/kg range x 10-8 m3/kg

topsoil 7 31.74 17.57 13.81 – 58.85 archaeological features 12 10.47 1.65 7.73 – 12.89 undisturbed subsoil (C-horizon) 6 9.63 1.54 7.61 – 11.09 all samples 25 16.22 13.29 7.61 – 58.85

Based on their morphology, the large, short wavelength magnetic anomalies encountered in both Harnaschpolder and Smokkelhoek could be identified as geological structures. Their appearance, in particular the ‘loop’ in the Smokkelhoek data, suggests a geological origin, for instance small creeks or gullies. They are most likely to be associated with either one of the two major marine transgression phases. A transect of borings was carried out at both sites in order to identify the features causing the magne-tic anomaly. Each of the cores was sampled for magnetic susceptibility measurements in the laboratory. The stratigraphy of the top four meters can be simplified to four layers; the silts of the Calais phase at the lower part of the sequence are overlain by Holland peat.

54

The peat, in turn, is covered by the silts and sands of the Dunkirk transgression phase, and in these silts a topsoil has developed. It must be noted that grey silts with black staining occur locally in sedi-ments of the Calais phase at both sites. Measurements of the magnetic susceptibility of the sediments in the laboratory showed that the stained Calais silts were at least partly responsible for the magnetic anomalies in the gradiometer survey. A possible contribution of any chemical or detrital remanent magnetism to the anomalies has not been investigated in this study. Table 9 presents the data of two cores, A and B, their location relative to the magnetic anomalies is indicated on Figures 23 and 24. Magnetic susceptibility values of the topsoil and the Dunkirk silts are generally very low. The lack of any clastic material in the peat explains the near-zero magnetic susceptibility values. One slightly negative value was caused by the presence of (diamagnetic) water in the samples, which were not dried prior to the measurements. The magnetic susceptibility of the grey layer with black stains in the Calais silts in these two cores is 5 to 100 times higher than the susceptibility of the Dunkirk silts. This is, however, a local phenomenon. Table 9 Magnetic susceptibility of samples from core A (Smokkelhoek) and core B (Harnaschpolder). depth in cm

material mass magnetic susceptibility x 10-8 m3/kg

above / below groundwater

core A 0 silty clay 6.6 above 60 silty clay + iron staining 5.9 above 120 peat 0.9 above 140 silty clay + organic material 5.0 below 180 peat -0.5 below 220 silty clay 58.9 below 240 silty clay 177.3 below 300 silty clay 35.2 below core B 10 silt 13.98 above 55 silt + iron staining 8.93 above 155 silty sand + iron staining 7.09 above 175 silt 4.64 below 210 sand 5.16 below 280 peat 0.99 below 310 silt 5.30 below 315 silt 199.56 below 319 silt 980.82 below Away from the magnetic anomalies, silts of the Calais phase have susceptibility figures similar to the Dunkirk silts. Microscopic inspection of the black stained soil showed that it contains pyrite crystals. The crystals are concentrated in and around macrofossils present in the soil. The samples were not further investigated, and only the presence of non-framboidal pyrite crystals was confirmed. Three subsamples of the Harnaschpolder core were measured before and after air-drying (Table 10). The upper sample (310 cm) has the lowest magnetic susceptibility of the three samples, and its susceptibility does not change upon oxidation. The magnetic susceptibility of the two lower samples (315 and 320 cm) however, decreased considerably after air drying. Oxidation of the samples apparently causes changes in the iron mineralogy that are reflected in the magnetic susceptibility of the samples. Table 10 Magnetic susceptibility of the subsamples from core A (Harnaschpolder) before and after air-drying. core depth (cm) material magnetic susceptibility

x 10-8 m3/kg before air-drying magnetic susceptibility x 10-8 m3/kg after air-drying

A 310 silt 6.02 6.02 315 silt 200.40 30.04 319 silt 981.34 40.79

55

5.3.5 Discussion The fluxgate gradiometer surveys in Harnaschpolder and Smokkelhoek failed to identify any archaeo-logical features. The data on the magnetic susceptibility of the archaeological features in Harnasch-polder, however limited in the number of samples, suggests that the magnetic contrast between the fill of the archaeological features and the matrix in which they are embedded is negligible. In general the values are very low compared to, for example, values found in the United Kingdom (e.g. Dearing et al. 1996). A heating experiment, however, showed that the soils could be magnetically enhanced. This indicates that it is not the lack of iron that suppresses the magnetic susceptibility in Harnaschpolder. Topsoil samples do have a higher magnetic susceptibility, as would be expected, but the values of the samples are not very consistent. This could be due to the presence of inclusions (the samples were not dried and not sieved), or to the fallout of industrial activities in the vicinity of the Harnaschpolder. Thus, archaeological features were not and possibly could not have been detected in the gradiometer survey in Harnaschpolder. The reason why can only be speculated upon in this stage of the research, and further investigations are required to shed light on the processes explaining this phenomenon. The following section will list some of the possible reasons for the lack of magnetic contrast in Harnaschpolder. These reasons can be split into two groups; either the magnetic contrast between the archaeological features and the natural matrix has never existed, or the contrast did exist but has been reduced over time. Into the first group falls the impediment of magnetic susceptibility enhancement by waterlogging, and the lack of vertical differentiation of the young soil. According to Weston (2002), waterlogging of a soil can prevent the enhancement of the magnetic susceptibility by heating. Linford and Canti (2001) came to similar conclusions in their experiments. Enhancement of a waterlogged clay soil was achieved by kindling a fire for a one and a four-day period. The enhancement is concentrated in the top four cm of the soil, but causes a clear magnetic anomaly. The lack of vertical differentiation is common in young sediments. Soil formation processes are hamperred by waterlogging of the soil, and because of ongoing marine transgression, the formation of a well-defined A-horizon is prevented. In the surroundings of Harnaschpolder, but not in the Harnaschpolder excavations, a vegetation level associated with the Roman Period habitation has been found locally. The appearance of the fill of the archaeological features on the Harnaschpolder exca-vations suggests that a topsoil with organic material has been present here, but post-Roman marine transgressions have probably eroded or reworked this layer. With the topsoil, the magnetic suscepti-bility variations contained in it would have been taken away, but this erosion should not have affected the fill of the archaeological features. The contrast between the fill of the archaeological features and the matrix they are embedded in is however very small. If the magnetic contrast was larger once, is it possible that it was reduced? Long term waterlogging of the soil could delete a magnetic enhancement. By continuous water-logging, ferrimagnetic iron oxides will dissolve, and they can eventually flush out of the soil profile (Thompson & Oldfield 1986). Short term waterlogging can cause iron compounds to change into either the ferri-magnetic maghemite, or to the non-ferrimagnetic green rusts or ferric or ferrous hydroxides upon oxidation (Weston 2002). Fractional conversion values indicate that the soils of Harnaschpolder were not flushed of iron, but chemical changes induced by waterlogging, may have changed the magnetic contrast. In sharp contrast to the lack of magnetic contrast between the fill of the archaeological features and the undisturbed matrix, strong features of a natural origin were detected in the magnetometer survey. In a magnetometer survey, induced and remanent magnetism cannot be distinguished from one another, and magnetic anomalies may be caused by either of the two, or a combination of both types of magnetization. Although a possible contribution of chemical or detrital remanence was not investigated, it can be inferred from the data that the magnetic anomalies that were encountered in both Harnaschpolder and Smokkelhoek were at least partly caused by bodies of highly magnetic material in the sediments of the Calais phase. Based on their black colour or staining and their geomorphological expression, these deposits are interpreted as anoxic shallow creek fills, in which stagnant water provided continuous reducing conditions. High magnetic susceptibility values have also been recorded just over or under the black layer. Pyrite was visually identified in these layers with the aid of a microscope.

56

Pyrite, however, is paramagnetic, and its magnetic susceptibility of 30 x 10-8 m3/kg cannot explain the magnetic anomalies encountered in Smokkelhoek en Harnaschpolder, and other minerals must be responsible. Although a full discussion falls outside the scope of this paper, a short review of iron sulphide formation and degradation is required to understand the phenomena encountered in this study. It is known that sedimentary iron sulphides occur in marine and estuarine environments (e.g. Luther et al. 1982, Rabenhorst 1990). The formation of these sulphides requires a source of ferrous iron (Fe(II)), a source of sulphur and a reducing environment. Fe(II) is present in sea water, but studies into the cycling of iron in salt marshes suggest that crystalline Fe(III) minerals from the sediment can be dissolved by organic ligands which makes them available for the formation of iron sulphides (Luther et al. 1982, Kostka & Luther 1995). Sulphates are abundant in sea water, and they can be reduced by sulphate reducing bacteria in the process of decomposition of organic matter. Thus iron sulphides can be formed in a reducing estuarine environment in the presence of organic matter. The end product could be one of the monosulphides mackinawite (FeS0.9) or greigite (Fe3S4), pyrrho-tite (~Fe7S8), or in fact pyrite (FeS2). Pyrite and mackinawite are paramagnetic, pyrrhotite and greigite however are ferrimagnetic, and the presence of either of these two iron sulphides, or a combination, could be responsible for the magnetic anomalies in Harnaschpolder and Smokkelhoek. Greigite oxidi-zes rapidly on exposure in air, and the data in table three may indeed indicate the presence of greigite in the waterlogged sediment. In order to fully understand the occurrence of iron minerals in estuarine sedimentary sequences it is essential to know the depositional, but also the post-depositional processes involved. For both sites, post-depositional processes can de separated into two phases. First, as part of the geological develop-ment of the landscape, a phase of drying out and oxidation takes place. After cessation of the marine influence the environment gradually becomes fresher. Lowering local groundwater levels and increasing vegetation lead to better aeration of the soil, oxidizing the iron sulphides to goethite (�-FeOOH) or haematite (�-Fe2O3), both paramagnetic minerals, or to ferrimagnetic magnetite (Fe3O4) (Luther et al. 1982) These newly formed iron compounds also occur in the now oxidized zone with archaeological features, and their distribution does not necessarily bare a relation to the distribution of iron compounds before the marine transgression. The second phase consists in both cases of renewed marine inundation of the landscape. Despite the abundant literature on the subject of sulphide formation, it is as yet unclear what the potential effect of saline waterlogging on the existing iron mineralogy of the soil is. Neo-formation of (primary) iron sulphides is likely to take place, but it can be envisaged that at least part of the iron oxide present will be reduced to sulphides again. In addition to the high magnetic susceptibility values already identified in this preliminary study, NRM may also have contributed to the magnetic anomalies identified. Evidence for the co-occurrence of high magnetic susceptibility and high NRM has been found in a freshwater environment by Ellis and Brown (1998). It is conceivable that processes similar to those postulated by Ellis and Brown have operated on the sites in the current study as well, but unfortunately the distinction between induced and remanent magnetism cannot be made from magnetometer survey data. Without further study therefore the contribution of NRM remains uncertain. 5.3.6 Conclusion This study has shown that estuarine (salt marsh) sediments pose several difficulties in the interpretation of magnetometer surveys carried out for archaeological purposes. Archaeological features may be indistinct or invisible, whereas geological structures may show with unexpected clarity. The reason for the lack of magnetic contrast between archaeological deposits and the un-disturbed soil matrix, is the topic of further research. Knowledge of the type of iron minerals present in the sites under investigation can lead to the understanding of the processes that hamper the creation of a contrast, or delete an existing contrast in magnetic susceptibility. (Magnetic) mineral investigation will also have to clarify whether it is indeed greigite or pyrrothite, or a combination of the two iron sulphide compounds, that is causing the magnetic anomalies encountered on the two sites discussed in this paper. The topic of the possible magnetic remanence of the sediments will also be addressed in further studies.

57

Marine transgressions can have a great influence on the iron mineralogy of the soil. It is likely that this would have affected iron oxides in the fills of archaeological features. Further research will be carried out on selected samples, in order to reconstruct the magnetic history of the investigated sites. 5.4 Iron compounds from archaeological features in estuarine deposits, examples from The

Netherlands7 5.4.1 Introduction Magnetometer surveys are probably one of the most successful methods of archaeological prospection (Clark 1996, Gaffney & Gater 2003), even to the extent that on occasions the term is used synony-mously with ‘geophysical surveys’ in general. Consequently, expectations in the reliability of map-ping subsurface anomalies are high, which can lead to disappointment if a magnetometer survey does not reveal all the buried archaeological features. Studies have hence been conducted to investigate the conditions under which magnetometer surveys are successful, or indeed how success can be measured (Hey & Lacey 2001). There are many reasons why archaeological features exhibit a magnetic contrast with the surrounding matrix that can lead to their detection in a magnetometer survey, but there are as least as many conditions under which no such contrast exists. Parameters influencing the results include geology, soil, site formation processes, post depositional processes, anthropogenic influences, site types, and many more. It is hence virtually impossible to compile an exhaustive list of all these conditions and studies into this subject have concentrated on selections of particular parameters. In recent years several authors have analysed the magnetic mineralogy of archaeological sites and linked results to site specific parameters (for example Fassbinder et al. 1990, Linford 1994, Crowther & Barker 1995, Weston, 1996, Ikeya et al. 1997, Marmet et al. 1999, Linford & Canti 2001, Weston 2002, Linford 2004, Weston 2004), confirming a research trend predicted by Dalan and Banerjee (1998). However, coastal and estuarine environments are notably absent from these studies despite their archaeological importance in countries including the United Kingdom, Germany and The Nether-lands. In a previous paper (Kattenberg & Aalbersberg 2004) the problems were highlighted that were encountered in magnetometer surveys over archaeological sites in Dutch estuarine soils. On a number of such sites in The Netherlands archaeological features could not be magnetically mapped, while broad, strongly magnetic anomalies with a creek-like appearance dominated the magnetic surveys, which would have masked weaker archaeological signals. This paper will address both issues by investigating the magnetic mineralogy of soils and sediments from two representative sites. 5.4.2 Background When magnetometer surveys were first tested in archaeology, it was anticipated that they would mainly detect anthropogenic structures with thermoremanent magnetisation, like kilns and hearths. However, it soon became apparent that features dug in the past (e.g. ditches, pits) which had become filled with soils and sediments over the centuries, can also produce clear magnetometer data (Clark, 1996). The reason is the contrast in magnetic susceptibility between the fill of the features and the surrounding soil (the ‘matrix’), which leads to differences in induced magnetisation that can be detected at the surface with a magnetometer. The following discussion therefore concentrates on the magnetic mineralogy of samples with a view to explain their magnetic susceptibility. Investigations of remanent magnetisation are not included. 5.4.2.1 Magnetic susceptibility Every material responds differently to the application of an external magnetic field, and for soils this is no different. Magnetic susceptibility measures the ease with which a material can be magnetized. Whether a deposit is natural or anthropogenic, its magnetic susceptibility depends mainly on the amount and type of the different iron compounds that it contains. Their concentration, grain size and grain shape all have an influence on the overall magnetic susceptibility.

7 Manuscript in preparation by A.E. Kattenberg, A. Schmidt, M.J. Dekkers, T.A.T. Mullender and H. Kars.

58

However, in first approximation, the magnetic susceptibility of soils is a measure of the concentration of magnetite and maghemite, since these iron oxides are by far the most magnetic compounds found in soils (see below). Every deposit is unique and is likely to have a magnetic susceptibility that is different from the deposits surrounding it. Intra-deposit variations are usually weak so that magnetic susceptibility measurements can sometimes be used to delineate archaeological contexts (Dockrill & Simpson 1994). It has been found that topsoil material and the fill of archaeological features generally have an enhanced magnetic susceptibility, compared to the surrounding matrix. This enhancement can often be related to the conversion of a non-ferrimagnetic into a ferrimagnetic iron compound. Before the underlying enhancement mecha-nisms can be discussed a brief description of the relevant iron oxides is required. In this section the iron oxides that are typical for temperate climate zones are summarised, using data from Taylor (1980), Weston (1999), Cornell and Schwertmann (2003), and Hansel et al. (2005). The chemical formula, colour and possible formation pathways of each iron oxide are briefly discussed, in addition to a description of well documented formation and transformation processes. However, the (trans)formation of iron in the soil is a complex process, which depends on many variables, including climate, soil pH, organic matter content and redox status of the soil. More advanced processes, not discussed here, are also possible and are best understood in the laboratory. In some cases the chemical formula for the iron oxides is the same (for example for hematite and maghemite) but the crystal structure is different, this is indicated with a prefix of � for the hexagonal and � for the cubic crystal structure. Goethite � FeOOH Goethite is the most common iron oxide in cool to temperate humid climates, where it usually coexists with lepidocrocite and ferrihydrite. It is yellowish brow in colour (7.5 to 10 YR). It can be formed from solid Fe(II) compounds like iron-carbonates or iron-sulphides, or reduced from Fe(III) by microbial action. Alternatively it can be transformed from ferrihydrite. Hematite � Fe2O3 Hematite is the second most common soil iron oxide, it mainly occurs in warmer to subtropical / tropical climates. It has a distinctively red colour (5YR to 10R). Hematite and goethite are closely related and can coexist, formation of either of the two iron oxides depends on temperature and drainage. Ferrihydrite, for example, can transform to goethite, but also to hematite in warmer and drier climatic conditions where the formation of hematite is preferred over the formation of goethite. In gley soils ferrihydrite may be the precursor of hematite. Lepidocrocite � FeOOH Lepidocrocite has been identified in different climatic zones, but not in calcareous soils. It is a very common iron oxide, usually orange in colour (5YR to 7.5YR). It is formed in seasonally wetting and drying (reductomorphic) environments, for example in iron pans. It can be transformed from ferrihydrite, or from magnetite after dissolution and consequent oxidation. Ferrihydrite 5Fe2O3•9H2O 5Fe2O3•9H2O is a possible formula for ferrihydrite, but other formulas have been proposed. Ferrihydrite is a hydrated ferric oxide, which can only be found in young (Holocene) deposits where its transformation to goethite or hematite has been impeded or delayed, or where circum-stances are detrimental to the formation of more crystalline iron oxides like goethite. These are environments where there is sufficient Fe(II) oxidation in the presence of organic matter and silicate. Ferrihydrite occurs in gley soils, on the oxidizing / reducing boundary of the soil and in the B-horizon of podzol soils. Magnetite Fe3O4 Lithoogenic magnetite occurs commonly in the coarse fraction of the soil. The abundance of magnetite in the topsoil when compared to the subsoil suggests a pedogenic formation. Magnetite is a reduced iron oxide, which may oxidize to form maghemite, turning in colour from black to brown. In an oxidizing environment, magnetite will usually oxidized partly, the resultant iron oxide will belong to the magnetite / maghemite series, but will neither be purely maghemite, nor purely magnetite. Magnetite can be formed trough the reduction of hematite and goethite, or the reaction of ferrihydrite with Fe(II).

59

Maghemite � Fe2O3 Maghemite is found mainly in tropical and subtropical regions, with localized deposits in temperate regions. The iron oxide occurs in concretions, or can be dispersed through the soil. It can be formed through the oxidation of magnetite, or the dehydration of lepidocrocite at a temperature of ~250 ºC. Taylor (1980) has synthesized maghemite from green rust during laboratory experiments, in a process that approximated the oxidation of a gley soil. The iron mineralogy of the aerobic part of the Dutch estuarine deposits that are investigated in this paper is, because of its calcareous nature and climatic conditions, expected to be dominated by ferri-hydrite and goethite, the former of which can, over time, transform into the latter. These two domin-ant iron compounds have a very low magnetic susceptibility. Hematite, however, is unlikely to be formed from ferrihydrite in our climatic zone, except possibly in gley soils. In topsoils the non-ferri-magnetic compounds are expected to be mixed with ferrimagnetic, and thus high magnetic suscepti-bility iron oxides of the magnetite / maghemite series. The main contribution to the magnetic suscepti-bility of the deposits will come from these ferrimagnetic minerals. 5.4.2.2 Enhancement of topsoil Induced magnetic anomalies caused by buried archaeological features can be positive or negative because the fill of cut archaeological features (e.g. ditches and pits) has a different – either higher or lower – magnetization than the matrix they are surrounded with. In practice, however, such buried archaeological features often cause positive anomalies. This is explained by the generally higher magnetic susceptibility of the topsoil with which these features have been filled after the sites were abandoned. The underlying processes for the enhancement of magnetic susceptibility of topsoil have been investi-gated by several authors, first by Le Borgne (1955) who proposed two pathways: enhancement by intense heating, and enhancement by fermentation. Since then other mechanisms were discovered and authors tried to modify Le Borgne’s terminology to encompass these pathways (Weston 2002). Authors have also often differentiated between primary and secondary ferrimagnetic iron oxides. The former are directly formed from the geological source (lithogenesis), while the latter are produced via other iron compounds as described above. All magnetic enhancement hence leads to secondary iron oxides. The following description of the five main pathways for magnetic enhance-ment is partly based on the terminology introduced by Linford (2004), which is congruent with Dalan & Banerjee (1998). In the first pathway, non-ferrimagnetic iron oxides like hematite, goethite and lepidocrocite, are converted into magnetite when heated under reducing circumstances in the presence of organic matter. The temperature at which this process starts is not well defined and values between 150 °C and 570 °C have been reported, with lower temperatures requiring longer exposure (Linford & Canti 2001, Maki et al. 2006). On cooling, this magnetite can oxidize to maghemite if sufficient organic matter is present. Both magnetite and maghemite are ferrimagnetic iron oxides with a high magnetic susceptibility. This first pathway is accepted as the most common form of enhancement in an archaeo-logical setting. The resulting magnetic susceptibility depends on the concentration of iron oxides initially available in the soil for conversion and the intensity of heating. The latter can be evaluated through measurement of the fractional conversion (see below) and is a rough indicator for the intensity of anthropogenic impact. The archaeological contexts in which such soil alterations may be found include hearths, kilns, furnaces and other high temperature features, but even bushfires can create enhanced magnetic susceptibility, although this may be partly attributable to the resulting ash layer (Peters & Thompson 1998, Linford & Canti 2001). It is likely that most of the enhanced material will gradually be spread over the surface, thus increasing the overall topsoil magnetic susceptibility. The second and third pathways require microbes, thriving in rich organic deposits, to reduce soil iron minerals (e.g. hematite) to their ferrimagnetic forms. ‘Microbially mediated’ (Linford 2004) conver-sion of the second pathway relies on the change of soil pH/Eh by microbes to create conditions that facilitate the creation of magnetite, either in close contact with the bacteria or in the general soil environment (Gibbs-Eggar et al. 1999). In the right climatic conditions (temperature, moisture) the subsequent re-oxidation of magnetite to maghemite can also be microbially mediated. In contrast to this passive extra-cellular process, magnetotactic bacteria actively create intra-cellular crystalline magnetite to navigate in the earth’s magnetic field (Fassbinder et al. 1990).

60

These fine-grained magnetite crystals remain in the soil after the bacteria die and form the third pathway of magnetic susceptibility enhancement. Since methanogenesis is not required for either of these biogenic pathways, Le Borgne’s term ‘fermentation’ is not strictly applicable (Weston 2002). However, it provides a useful connotation for the organic matter that microbes need to grow. Archaeological environments with rich organic deposits include, for example, middens and decayed wooden posts. Fassbinder demonstrated that magnetic anomalies of post holes that were apparent in high-sensitivity magnetometer surveys are attributable to the remains from magnetotactic bacteria (Fassbinder & Stanjek 1993, Fassbinder & Irlinger 1994). A fourth pathway for the magnetic enhancement of topsoil is the addition of magnetic material, for example broken pottery or brick fragments (Weston 2002). Such material is often found as discard or rubbish in archaeological middens and has been spread on arable fields with other manure, mainly in Medieval times. Metalworking remains, for example hammerscale and slag, also become incorporated into soil layers and can greatly increase the magnetisation. Iron and steel fragments broken or fallen from modern farming machinery can also enhance the magnetic make-up of the topsoil, but often also create undesirable isolated magnetic anomalies. Enhancement of soil magnetic susceptibility also occurs during soil formation processes (pedo-genesis). Maher and Taylor (1988) reported the formation of ultra-fine grained magnetite in soil despite the absence of any microorganisms. This is considered the fifth pathway of enhancement. The first three pathways rely on the availability of organic matter, which is usually more abundant in the upper soil horizon than in the subsoil, hence creating a magnetic differentiation of topsoil and subsoil. In addition, anthropogenic input further enhances these conditions (either through fire or deposition of organic material, like middens) sometimes allowing the identification of settlement areas through magnetic susceptibility mapping, or the differentiation of buried land surfaces (e.g. covered by windblown or alluvial deposits) from the magnetic stratigraphy. All enhancement pathways lead to the presence of a greater abundance of magnetite, maghemite, or a combination of these two ferrimagnetic iron oxides in the surface soil, in the fill of archaeological features or in the wooden structures on archaeological sites. Whenever a cut archaeological feature is filled with such enhanced soil the magnetic susceptibility contrast with the surrounding soil or sediment matrix will lead to induced magnetisation which produces a magnetic field measurable at the surface. However, archaeological features do not always show a positive magnetic contrast. Linford (1994) and Weston (2002) have tried to explain the lack of a distinguishable magnetometer signal on two British archaeological sites, and Maki et al. (2006) investigated the origins of a negative magnetic anomaly that was caused by a series of archaeological hearths. Similarly, no magnetic contrast was found on some archaeological sites under estuarine deposits in The Netherlands. Two were discussed in an earlier paper (Kattenberg & Aalbersberg 2004) and two more examples are investigated in this study. There are two possible explanations for the observed lack of magnetic contrast. Either the contrast has never existed (or was extremely weak), or subsequent processes have altered the iron mineralogy of the soil and thereby changed the magnetic susceptibility. Soil is a dynamic medium, and chemical changes can be caused for example by wetting and drying, or by the persistent water-logging of a soil. A detailed study of the soil iron compounds of archaeological deposits in estuarine soils in the Netherlands was hence undertaken. 5.4.3 Methods and materials Investigated sites Two specific sites were selected for this study as they most clearly display the problems of insuffi-cient magnetic contrast in archaeological features despite good potential for enhancement in the matrix. The sites of Broekpolder and Harnaschpolder (Fig. 25) developed in an estuarine environment and were, after their abandonment, subject to renewed marine transgression and seawater or brackish water logging, with associated blanketing through estuarine deposits. Broekpolder The Bronze Age / Iron Age settlement of Broekpolder is a scheduled archaeological monument. Exca-vations in the area around the current monument showed many archaeological features connected to habitation, agriculture and religion ranging from the Bronze Age to the Middle Ages (Terkorn et al. in press).

61

Figure 25 The extent of marine and estuarine deposits in The Netherlands. B = Broekpolder, H = Harnasch-polder.

The area under protection was only investigated with test trenches. In this latter area a magnetometer survey was carried out, in which none of the confirmed archaeological features (mainly pits and ditches) could be magnetically mapped (Fig. 26). A modern ditch, however, did have a disting-uishable magnetic response. In the section of a drainage trench that was freshly cut during the survey a prehistoric ditch could be identified. Harnaschpolder At Harnaschpolder an Iron Age / Roman Period settlement with an associated parcelling system was investigated with a combination of test trenches and excavations (Flamman & Besselsen, in press). A magnetometer survey of part of the off-site area (Fig. 27) did not locate the continuation of any of the Roman Period ditches or indeed of any of the other archaeological features that were confirmed by excavation, but a post Medieval ditch could be mapped. In addition to strong anomalies caused by modern ferrous pipes, highly magnetic anomalies of a possible geological nature (Fig. 27), not dissi-milar to the features encountered in Smokkelhoek (Kattenberg & Aalbersberg 2004), were mapped in the NE part of the survey area. During the excavation of an associated Roman Period settlement site, due south of the surveyed area, samples could be collected for investigation. To analyse the iron mineralogy of the soil, samples from archaeological features as well as from undisturbed soil were taken and several methods of magnetic investigation were used, details of which are briefly described in the following sections. Low frequency magnetic susceptibility measurements For bulk soil samples the magnetic susceptibility was measured on an AC magnetic susceptibility bridge in the Department of Archaeological Sciences at the University of Bradford, UK. For calibration, samples of manganese sulphate and high alumina cement were used to derive mass specific magnetic susceptibility values. For each sample averages of three instrument readings and two weight determinations were calculated. For small sample masses (e.g. in advance of IRM measurements) magnetic susceptibility was measured on an Agico KLY-2 Kappabridge in the Palaeomagnetic Laboratory of the University of Utrecht, The Netherlands. Fractional conversion Fractional conversion was determined according to a procedure similar to that described by Clark (1996). Approximately 10 ml of each sample was selected as a subsample (see Tables 13 and 14) and weighed on a top-pan balance.

62

Figure 26 The site of Broekpolder with the results of the magnetometer survey, the location of the magnetic susceptibility samples (black dots) and the fractional conversion samples (black dots with core numbers). Also shown are ditch section A (Fig. 28) and the location of the modern ditch, where samples for IRM measurements were collected. The location of the trial trenches and the archaeological features within them has been super-imposed on the results of the magnetometer survey. Large magnetic anomalies have been caused by dipwells that were installed on the monument to monitor groundwater level and other parameters (Van Heeringen et al. 2004).

63

Figure 27 The site of Harnaschpolder with the results of the magnetometer survey and the location of the cores from which samples were collected for magnetic susceptibility measurements (grey), for fractional conversion measurements (black) and for thermomagnetic measurements (core PYR). The data of fractional conversion samples of cores A60 and A180 are presented in this paper (Table 14). The section from which the IRM measurements are taken is located 400 meter to the south of the displayed area. The volume of the samples was measured using a 25 ml cylinder and its magnetic susceptibility determined as described above. One gram of plain flour was added to each sample before it was placed in a porcelain crucible and covered with a porcelain lid. A Carbolite electric muffle kiln was heated up to 650 ºC with the chimney closed. Once this temperature was reached the samples were placed in the furnace and left in the furnace for one hour after the kiln had reached a temperature of 650 ºC again. This procedure creates high temperatures and reducing conditions sufficient to convert most antiferromagnetic iron oxides to magnetite. The furnace was then switched off and left to cool for half an hour before the samples were taken out. The lids were removed and the samples were stirred with a wooden spatula. The furnace was switched on again with the chimney open. After reaching 650 ºC the samples were put back into the furnace and heated for 45 minutes, achieving oxidising conditions with organic matter in close contact with the soil. The furnace was then switched off and left to cool for half an hour before taking the samples out. After reaching room temperature, the volume of the samples was measured using a 25 ml cylinder and they were weighed using a top-pan balance. The magnetic susceptibility of the samples was measured as described above. The ‘fractional con-version’ of a sample is calculated as the ratio of its initial magnetic susceptibility to the value after the heating cycle. Since magnetic susceptibility of soil samples is dominated by ferrimagnetic iron-oxides, the initial measurement indicates how much was created in antiquity while the latter readings show how much could have been converted overall. Fractional conversion is hence a rough indicator for the intensity of conversion in the past. IRM components Magnetic susceptibility of the samples was measured with a KLY-2 Kappabridge (see above) before they were demagnetised with an AF demagnetiser.

64

Using a PM4 pulse magnetizer, Isothermal Remanent Magnetisation (IRM) was acquired in the samples in steps of increasing field strength: 0, 10, 15, 20, 25, 30, 40, 50, 65, 80, 100, 120, 150, 180, 200, 250, 300, 400, 500, 650, 800, 1000, 1200, 1400 and 1600 mT. The samples were measured with a JR-5A spinner magnetometer after every step. A component analysis was carried out on the results with the IRM-CLG 1.0 program developed by the Palaeo-magnetic Laboratory of the University of Utrecht (Kruiver et al. 2001, Heslop et al. 2002) and two components were fitted to the IRM curves since over-fitting with three or more components can produce ambiguous data. This analysis is based on a comparison of IRM curves and allows to distinguish components based on their magnetic hardness. IRM plots are a linear superposition of cumulative log-normal curves for each magnetic constituent, described by three parameters (Kruiver et al. 2001): the saturation magnetisation (SIRM), the field at which half the SIRM is reached (B1/2), which is an indicator for the magnetic hardness, and the width of the distribution (dispersion parameter DP), given by one standard deviation of the logarithmic distribution. The relative amplitude (SIRM) of each component’s distribution can also be expressed as a percentage of the whole fit. In order to convert these to elemental concentrations, typical SIRM values for the components would have to be used. The width of the distribution (DP) is often used as a measure for the crystallinity of a sample (high crystallinity for low DP). For the interpretation of the acquired data the following B1/2 values have been used to distinguish components: soft component (ferrimagnetic) 30-50 mT magnetite (with low DP) 50-60 mT oxidized magnetite (maghemite) (with high DP) hard component (anti-ferromagnetic) 0.5 to 1.5 T hematite 1.5 to 2 T goethite These boundaries are based on values published by Kruiver et al. (2001, 2003), Kruiver & Passier (2001) and Heslop et al. (2002). Thermomagnetic measurements To further analyse the magnetic composition of the soils a modified horizontal-translation-type Curie Balance was used in the Palaeomagnetic Laboratory of the University of Utrecht, The Netherlands (Mullender et al. 1993). This balance, with a sensitivity of approximately 5 x 10-9 Am2, has been modified to enable the separation of the non-ferrimagnetic contribution (SIG2) to the total magnetisation (SIG1). By cycling the applied field between 150-300 mT, the ferrimagnetic signal is saturated, whereas the non-ferrimagnetic part of the signal has a linear dependency on the applied field. SIG1 is equivalent to the classical thermomagnetic signal that is obtained in a steady field. For this study only the total magnetization, calculated form both SIG1 and SIG2, is used. Samples were placed in a quartz glass sample holder, fixed with quartz glass wool, and heated and cooled in air in 16 runs of 15-150, 150-50, 50-250, 250-150, 150-300, 300-200, 200-350, 350-250, 250-400, 400-300, 300-500, 500-400, 400-600, 600-500, 500-650, 650-15 °C, at a heating rate of 10 °C min-1 and a cooling rate of 15 °C min-1. This analysis provided further distinction of iron compounds, based on their thermal behaviour. In particular, it allowed the identification of iron sulphides (Dekkers et al. 2000). Samples Broekpolder In the area of the magnetometer survey of Broekpolder, samples for magnetic susceptibility measure-ments were taken by hand auger (Fig. 26) from the topsoil, the undisturbed soil matrix, the fills of possible archaeological features, and the 'presumed archaeological level', which was not recognized in the cores, but is known to be present at the interface between the topsoil and the undisturbed matrix. Samples were taken from the auger with a spatula and stored in polyester ziplock bags. They were then air-dried on paper plates, and ground with a porcelain mortar and pestle. After the magnetic susceptibility measurements, nine samples were selected for fractional conversion measurements (Fig. 26).

65

sam

ple

mag

netic

su

scep

tibili

ty x

10

-8 m

3 /kg

mas

s x

10-3

kg

desc

ript

ion

com

pone

nt

B1/

2 mT

SIR

M A

/m

SIR

M x

10 -3

A

m2 /k

g

SIR

M/�

A/m

DP

inte

rpre

tatio

n

cont

ri-b

utio

n to

si

gnal

%

1 18.68 0.50 topsoil 1 42.2 0.78 15.6 83511.78 0.39 magnetite 93 2 631.0 0.055 1.1 5888.651 0.22 hematite 7 2 5.84 0.52 undisturbed 1 56.2 0.137 2.634 45102.74 0.46 oxidized magnetite 91 2 707.9 0.014 0.2692 4609.589 0.30 hematite 9 3 6.22 0.50 undisturbed 1 56.2 0.132 2.64 42443.73 0.40 oxidized magnetite 86 2 707.9 0.022 0.44 7073.955 0.37 hematite 14 4 21.13 0.47 topsoil 1 42.7 0.88 18.723 88608.61 0.35 magnetite 91 2 501.2 0.085 1.8085 8558.921 0.47 hematite 9 5 4.83 0.49 feature 1 53.7 0.124 2.5306 52393.37 0.42 oxidized magnetite 94 2 501.2 0.0085 0.1735 3592.133 0.30 hematite 6 6 6.78 0.51 feature 1 52.5 0.106 2.0784 30654.87 0.38 oxidized magnetite 91 2 501.2 0.0105 0.2059 3036.873 0.48 hematite 9 7 10.55 0.50 modern feature

(not in picture) 1 41.7 0.555 11.1 105213.27 0.38 magnetite 99

2 631.0 0.03 0.6 5687.20 0.2 hematite 1 Figure 28 The sampling location of the Broekpolder samples that were collected in ditch section A (top) and the results of the magnetic susceptibility measurements and the component analysis of the IRM data for these samples. The prehistoric ditch, which was identified in a freshly cut drainage trench (Section A, Fig. 28), was sampled for IRM measurements (upper and lower fill). Small plastic sample tubes were pushed into the exposed section and also used for transport and storage. As the samples were obtained from above the water table (aerobic part of the section), they were not freeze- but air-dried. Similar samples were taken from the topsoil, the undisturbed matrix and from the modern ditch fill (the latter by hand auger), which had produced the clear anomaly in the magnetometer data.

66

Samples Harnaschpolder At Harnaschpolder, a series of cores (three cm gouge auger) was established through the centre of the surveyed area (Fig. 27) in order to investigate the magnetic susceptibility and the potential for magnetic susceptibility enhancement (fractional conversion). From the cores, samples from the topsoil, the presumed archaeological level, the undisturbed material just below the archaeological level and from the undisturbed matrix at approximately 1 m depth were collected.

sam

ple

mag

netic

su

scep

tibili

ty

x 10

-8 m

3 /kg

mas

s x

10-3

kg

desc

ript

ion

com

pone

nt

B1/

2 mT

SIR

M A

/m

SIR

M x

10 -3

A

m2 /k

g

SIR

M/�

A/m

DP

inte

rpre

tatio

n

cont

ribut

ion

to

sign

al %

9 18.69 0.48 topsoil 1 38.9 0.95 19.7917 105894.418 0.39 magnetite 91 2 1995.3 0.089 1.8542 9920.63492 0.43 goethite 9 10 12.69 0.52 upper fill 1 40.7 0.218 4.1923 33036.3096 0.44 magnetite 90 2 1258.9 0.023 0.4423 3485.48221 0.45 hematite 10 11 12.89 0.49 lower fill 1 43.7 0.119 2.4286 18840.7403 0.45 magnetite 94 2 501.2 0.008 0.1633 1266.60439 0.4 hematite 6 12 11.03 0.53 undisturbed 1 50.1 0.139 2.6226 23777.3482 0.44 oxidized magnetite 89 2 794.3 0.017 0.3208 2908.02101 0.45 hematite 11

Figure 29 The sampling location of the Harnaschpolder samples that were collected from a Roman Period ditch (top) and the results of the magnetic susceptibility measurements and the component analysis of the IRM data (bottom). The topsoil sample has been taken from the section that is visible in the background of the photo. Again, the archaeological layer was not always recognized during coring, but the Iron Age / Roman Period surface was known to be present at the interface of the topsoil and the undisturbed matrix. In two cores a possible old land surface (A-horizon) could be identified. Twelve of the samples were selected for the heating experiment and subsequent measurement of their fractional conversion, representing the topsoil, the A-horizon, the presumed archaeological layer, the fill of a possible archaeological feature and the undisturbed soil matrix. Six samples, from cores A60 and A180 (Fig. 27) are presented in this paper. A series of samples was taken from a section of one of the Roman ditches that was exposed during the excavation south of the surveyed area (Fig. 29).

67

The topsoil, the upper fill, the lower fill and the adjacent undisturbed matrix were sampled by pushing plastic tubes into the exposed section (see above) and investigated with IRM measurements. The highly magnetic linear anomaly just east of the centre of the survey area (Fig. 28) was sampled with core PYR to be thermomagnetically measured on a Curie balance. A seven cm Dutch (screw)-auger was used for the first meter and a three cm gouge auger for samples deeper than a meter. Since some samples were taken from below the groundwater table (anaerobic part of the section), they were stored in lidded plastic tubes and freeze-dried in order to avoid any possible chemical changes upon exposure to air. To establish whether this treatment had an influence on the magnetic composition of the soil, the magnetic susceptibility of a second set of 46 samples from a site in a similar setting as Harnaschpolder (Smokkelhoek, see Kattenberg and Aalbersberg (2004) for details), was measured before and after the freeze-drying using a KLY-2 susceptibility bridge (see above). Only minor changes in the magnetic susceptibility of the samples were observed. Fourteen samples did not change at all, 16 samples changed less than 5% and 9 samples less than 10%. The remaining four samples changed by more than 10%., but in absolute terms this was less than 1 x 10-8 m3/kg in all cases. It can therefore be assumed that limited chemical reactions had taken place in the samples during the process of freeze-drying. This paper reports on three samples, from a depth of 0.05-0.15 m (above the groundwater table), 3.74-3.76 m and 3.76-3.78 m (below the groundwater table). 5.4.4 Results Magnetic susceptibility measurements The results of the magnetic susceptibility measurements on the Broekpolder samples are summarized in Table 11. The values for the magnetic susceptibility of all the samples are rather low, when compared to, for example, England (see discussion). There is a magnetic contrast, however, between the topsoil and the undisturbed matrix, and between the fill of the presumed archaeological features and the undisturbed matrix. Table 11 A summary of the magnetic susceptibility values of samples from Broekpolder. N is number of samples.

mean magnetic susceptibility x 10-8 m3/kg

standard deviation x 10-8 m3/kg

range x 10-8 m3/kg N

topsoil 17.55 2.38 14.65 - 21.68 12 ‘archaeological level’, under topsoil 5.6 0.76 4.82 - 6.66 5 ?archaeological feature 10.47 4.67 7.02 - 21.90 10 undisturbed 4.98 1.72 2.98 - 9.80 19 Table 12 A summary of the magnetic susceptibility values of samples from Harnaschpolder. N is Number of samples. Samples that deviate from the mean by more than 5 standard deviations have been excluded; these are topsoil A140 (181.86 x 10-8 m3/kg) and archaeological level A0 (40.16 x 10-8 m3/kg). These samples are assumed to have very magnetic inclusions like metal, brick or hard coal.



Range x 10-8 m3/kg N

topsoil 12.88 2.69 8.65 - 19.30 14 ‘archaeological level’, under topsoil 9.81 1.83 6.79 - 12.60 12 undisturbed 7.05 1.67 3.28 - 9.03 11 undisturbed 100 cm depth 5.19 2.41 0.48 - 9.23 15 Results from the auger transect in Harnaschpolder (Table 12) show a decrease of magnetic susceptibility with depth in all cores for deposits above the groundwater table. Magnetic susceptibility values are low, and in a similar range to the Broekpolder samples. There is a contrast between the topsoil and the undisturbed matrix, but no archaeological feature fills have been sampled. On neither of the sites an enhancement could be observed in samples taken from the presumed archaeological level. Fractional conversion The results of the fractional conversion measurements of the samples from Broekpolder are displayed in Table 6 and show low values throughout.

68

This is comparable to the values found by Crowther and Barker (1995) for cretaceous clays in Luton (UK) which had a fractional conversion of 0-4%, and floodplain / levee deposits in Derby (UK) with fractional conversion values ranging between 0 and 5 %. Low values reflect the low initial magnetic susceptibility of the samples in combination with the much-increased susceptibility after heating. The samples are sorted by their susceptibility after ignition, which gives a rough estimate of the amount of iron that is present in the sample before ignition. Very high values are obtained in the samples from the archaeological level, the archaeological features and the modern ditch fill. Lower magnetic susceptibility is observed in samples from the topsoil and the undisturbed topsoil. The iron, in non-ferrimagnetic form, is concentrated in the deposits between the topsoil and the undisturbed matrix (i.e. in the presumed archaeological layer and within the possible archaeological features). The results of the heating experiment for samples from Harnaschpolder are illustrated in Table 14 with two representative core profiles. The fractional conversion percentages are higher when compared to Broekpolder, indicating that the relative amount of already converted iron oxides before ignition was greater than at Broekpolder. In core A60, on the east side of the survey area, both the initial magnetic susceptibility and the acquired susceptibility decrease down the profile. Core A180, on the west side of the area, shows a markedly different result. Whereas the initial magnetic susceptibility decreases down the profile, after ignition the possible A-horizon sample, just below the topsoil, has the highest susceptibility, a similar situation to Broekpolder. Table 13 The results of the heating experiment for samples from Broekpolder. Samples are sorted by des-cending magnetic susceptibility after ignition. core interval

(cm) magnetic susceptibilty x 10-8 m3/kg

material interpretation magnetic susceptibility subsample x 10-8 m3/kg

magnetic musceptibility after ignition x 10-8 m3/kg

fractional conversion %

A140 25-35 5.14 grey sandy clay with iron staining

‘archaeological level’, under topsoil

4.46 1053.5 0.42

B40 30-40 6.17 mixed brown grey sand

?archaeological feature

6.11 987.58 0.62

A100 20-30 5.30 light brown grey sandy clay with iron staining

‘archaeological level’, under topsoil

5.27 984.91 0.54

A160 25-35 7.92 grey sandy clay with iron staining

modern ditch fill

7.89 889.13 0.89

A180 15-45 7.93 dark grey sandy clay with iron stains and organic matter

?archaeological feature

7.03 679.13 1.04

A80 40-50 4.20 light brown grey clayey sand with iron staining

undisturbed 3.44 671.34 0.51

B20 5-15 17.37 brown grey clayey sand

topsoil 16.49 498.88 3.31

A80 5-20 16.96 brown sandy clay topsoil 14.81 456.96 3.24 A80 85-90 2.98 grey sand with

iron staining and shell

undisturbed 2.52 119 2.12

IRM components The results of the IRM component analysis are displayed in Figures 28 and 29. For all samples the two identified components can be described as: (i) a magnetically soft, ferrimagnetic component, either in the form of magnetite or oxidized

magnetite (approaching maghemite) (ii) a magnetically hard anti-ferromagnetic component, either as hematite or goethite.

69

Table 14 Cores A60 and A180 (see Fig. 27 for location) with the results of the heating experiment. core top

of layer (cm)

magnetic susceptibilty x 10-8 m3/kg

material interpretation magnetic susceptibility subsample x 10-8 m3/kg

magnetic susceptibility after ignition x 10-8 m3/kg

fractional conversion %

A60 0 13.6 brown grey clayey silt topsoil 14.8 288.21 4.92 30 10.24 brown grey clayey silt 'archaeological

level' 7.95 221.57 3.59

40 7.67 light brown grey sandy silt with sandy layers and shell

undisturbed - - -

80 - light brown grey sandy silt with sandy layers and shell and iron staining

undisturbed - - -

150# - - - - - - A180 0 13.85 grey brown clayey silt topsoil 13.5 376.77 3.58 35 12.52 brown grey silty clay ?A-horizon 11.56 453.96 2.55 45 7.49 light brown grey silty

clay with sandy layers and iron staining

6.65 114.23 5.82

60 6.3 light brown grey very silty clay with sandy layers and iron staining

- - - -

100# - - - - -

In Broekpolder (Fig. 28), the only hard component that was identified in the ditch section samples is hematite. Therefore, there are only two iron oxide combinations; magnetite with hematite and oxi-dized magnetite with hematite. Magnetite is present in the two topsoil samples and in the modern ditch fill, whereas oxidized magnetite occurs in the fill of the archaeological feature and in the un-disturbed matrix. Based on iron mineralogy, the modern ditch has a topsoil-like fill, the archaeo-logical ditch has not. There is more variation in the Harnaschpolder samples (Fig. 29). In our interpretation, the topsoil and the upper fill of the ditch (i.e. the latest fill) have a similar iron mineralogy, containing both magnetite and goethite. The lower (primary) fill of the Roman Period ditch contains magnetite as well, but alongside hematite. The sample of the undisturbed matrix resembles the undisturbed and archaeo-logical feature samples from Broekpolder with oxidized magnetite as the soft component and hematite as the hard component. Thermomagnetic measurements In Harnaschpolder, core PYR (Fig. 27) was obtained for thermomagnetic measurements. The results of a representative set of three of the samples are displayed in Figure 30. These three graphs represent the following samples: (a) 5-15 cm, low magnetic susceptibility (17 x 10-8 m3/kg), above the groundwater table (Fig. 30a); (b) 374-376 cm, low magnetic susceptibility (15 x 10-8 m3/kg), below the groundwater table (Fig.

30b); (c) 376-378 cm, high magnetic susceptibility (114 x 10-8 m3/kg), below the groundwater table (Fig.

30c). In the Curie plot of Figure 30a there is no clear evidence for the presence of a certain type of ferri-magnetic iron compound in the sample. The hyperbolic (and reversible up to ~300 °C) shape of the plot is an indication of a mostly paramagnetic behaviour of the sample, probably caused by the clastic material that forms the bulk of the sample. The irreversible decrease of the total magnetization of the sample after the last heating run, however, is an indication that before heating there was some ferrimagnetic material present in the samples, which was converted into a less magnetic compound by the heating. In the lower temperature part of Figure 30b there is an irreversible decrease of the total magnetization from room temperature to approximately 315 °C.

70

HP02 core PYR 5-15 cm

0.00E+00

5.00E-03

1.00E-02

1.50E-02

2.00E-02

2.50E-02

3.00E-02

3.50E-02

0 100 200 300 400 500 600 700

temperature ˚C

tota

l mag

net

izat

ion

Am

2 /kg

Figure 30a Results of the thermomagnetic measurements on samples from core PYR, heating in a solid line, cooling in a dashed line; 5-15 cm, low magnetic susceptibility, above the groundwater table. The Curie plot shows mainly paramagnetic behaviour with a small ferrimagnetic component.


0.00E+00

1.00E-02

2.00E-02

3.00E-02

4.00E-02

5.00E-02

6.00E-02

7.00E-02

8.00E-02

9.00E-02

1.00E-01

0 100 200 300 400 500 600 700

temperature °C

tota

l mag

netiz

atio

n A

m2 /k

g

Figure 30b As Figure 30a; 374-376 cm, low magnetic susceptibility, below the groundwater table. The thermomagnetic behaviour of greigite (Fe3S4) has been described by Dekkers et al. (2000), who state that greigite shows such typical irreversible decrease of magnetization between ~250 °C and ~350 °C. A Curie temperature for greigite is difficult to establish due to the thermal decomposition of the compound that occurs already at low temperatures. In the higher temperature range the plot is dominated by the neo-formation of a ferrimagnetic compound above the temperature of 400 °C.

71


0.00E+00

2.00E-02

4.00E-02

6.00E-02

8.00E-02

1.00E-01

1.20E-01

1.40E-01

1.60E-01

0 100 200 300 400 500 600 700

temperature ˚C

tota

l mag

netiz

atio

n A

m2 /k

g

Fig 30c As Figure 30a; 376-378 cm, high magnetic susceptibility, below the groundwater table. The irreversible decrease of magnetization up to 315 °C may indicate the presence of greigite. Due to the irreversible decrease of the total magnetization from 500 °C up to 580 °C, the newly formed compound can most likely be identified as magnetite (Curie temperature of 580 °C), although the total magnetization and the apparent Curie temperature may be influenced by the chemical changes that have taken place in the sample during the measurement cycle. Greigite that is heated under air will form magnetite starting from 400 °C in natural samples (Dekkers et al. 2000) and the oxidation of pyrite to magnetite starts at approximately 420 °C (Van Velzen & Zijderveld 1992). Given the low magnetic susceptibility of the sample, only a small amount of greigite is expected to be present. The neoformation of magnetite may in this case mainly be caused by the oxidation of pyrite in combination with the oxidative alteration of a small amount of greigite. The Curie plot in Figure 30c is dominated by the irreversible drop in magnetization at lower tempe-rature ranges, from room temperature to 320 °C. As in sample (b), this may indicate the presence of greigite in the sample. Neo-formation of a ferrimagnetic compound starts at approximately 410 °C. After a peak in magnetization, the total magnetization decreases irreversibly to a temperature of 620 °C, suggesting that maghemite (Curie temperature approximately 600-675 °C) rather than magnetite (Curie temperature of 580 °C) may be the final ferrimagnetic product of this thermomagnetic run. However, due to chemical alterations in the sample during the thermomagnetic investigations, this conclusion remains speculative. The contributions of greigite and/or pyrite to this newly formed com-pound cannot be separated. The high magnetic susceptibility of the sample suggests that the ratio of greigite to pyrite is higher in this sample than in sample (b). This also explains why the initial irreversible decrease of magnetisation is more pronounced in this sample compared to (b). 5.4.5 Discussion The magnetic susceptibility of the topsoil and the subsoil layers at Harnaschpolder and Broekpolder is very low when it is compared to, for example, data from England, where a set of topsoil samples has been collected in 10 km x 10 km grid squares, and measured for magnetic susceptibility (Dearing et al. 1996, 2001). The average topsoil magnetic susceptibility of the 1176 English samples that were collected was found to be 73 x 10-8 m3/kg. This set includes soils that contain primary ferrimagnetic minerals (i.e. created in lithogenesis), whereas the estuarine soils under investigation in this paper are expected to consist mainly of secondary ferrimagnetic minerals (i.e. derived from other iron com-pounds), which makes it difficult to compare the two sets of data.

72

However, even when the samples are differentiated into categories of different geological back-ground, the mean magnetic susceptibilities of the English samples are high in comparison to the Dutch samples, the lowest being an alluvial gley soil with a magnetic susceptibility of 45 x 10-8 m3/kg, and a typical stanogley with a magnetic susceptibility of 36 x 10-8 m3/kg, compared to typical values of 4-20 x 10-8 m3/kg in this investigation. Not only were the magnetic susceptibility values low, the samples from the archaeological deposits and the undisturbed matrix showed a limited magnetic contrast between them. The lack of archaeo-logically caused magnetic anomalies in the data from the magnetometer survey was confirmed by the magnetic susceptibility measurements. Samples from cores as well as those from the sections in Broekpolder and Harnaschpolder showed that there was no major magnetic susceptibility contrast between the undisturbed matrix and the suspected archaeological level (and fill) of archaeological features. Notably, however, a small enhancement of the archaeological feature fills was observed in the core samples from Broekpolder. Would this contrast be sufficient for an archaeological feature to be detected in a magnetometer survey? Taking this latter contrast (5 x 10-8 m3/kg) as an example, a typical pit with dimensions 0.75 m x 0.75 m x 0.5 m, which is buried at 0.5 m depth would cause an anomaly of ca. 1 nT in a single sensor magnetometer, and slightly less in a gradiometer. Although within the detection limits of the fluxgate gradiometer that was used in this study, the anomaly is small and may be difficult to detect if there is any soil noise. Moreover, there is no consistent magnetic susceptibility contrast (i.e. either always positive or always negative) in the archaeological samples, or not even within the same feature. In the Broekpolder ditch section, for example, the upper fill has a negative contrast to its matrix, whereas the primary fill has a positive contrast. It is hence possible that overall no noticeable magnetic anomaly is created. Apart from the limited magnetic susceptibility contrast between the archaeological deposits and the undisturbed matrix in Broekpolder, the IRM data showed that the deposits were mineralogically similar as well. Whereas the undisturbed matrix and the feature fill samples contain a combination of oxidized magnetite and hematite, the topsoil and the modern feature fill contain magnetite and hematite. It is hence likely that the fill of the modern feature consists mainly of topsoil material, which makes it detectable in a magnetometer survey because of the contrast to the undisturbed matrix. Question remains why the undisturbed matrix and the archaeological feature fills are similar? Based on the results of the heating experiment, the soils do appear to have a good potential for magnetic susceptibility enhancement. In Broekpolder, the magnetic susceptibility of the topsoil is slightly enhanced when compared to the undisturbed matrix, but curiously, the topsoil samples have much lower magnetic susceptibility values after ignition than the presumed archaeological level and the fill of modern and archaeological features. The convertible iron therefore appears to be concentrated in the deposits at the interface of the topsoil and the undisturbed matrix, but, judging from the low magnetic susceptibility values, largely in a non-ferrimagnetic form. The topsoil samples contain less iron, but more of it is already converted into a ferrimagnetic form. There are two ways to explain the similarity, both in magnetic susceptibility and in iron mineralogy, of the fill of the archaeological feature and the undisturbed subsoil, either the two deposits were already similar at the time of deposition, or one or both of the deposits have changed due to post depositional processes. A similarity between the undisturbed matrix and the feature fill could be due to the fill being derived from the undisturbed deposits themselves (e.g. caving in of the sides of the feature) or if it is filled with an undeveloped topsoil. This is unlikely because of the distinct colour difference between the archaeological feature fill and the matrix, the former being darker in colour and containing organic matter, i.e. being ‘topsoil-like’. The modern feature appears to be filled with topsoil material. Because of its appearance it is possible that the archaeological feature is filled with (ancient) topsoil-like material. Possibly a similar contrast as in the modern feature was once present in the fill of the archaeological feature, but has disappeared under the influence of post depositional processes. The processes that may have been of importance for the sites under investigation, the dissolution of iron oxides by gleying, leaching, water logging and seawater logging are discussed in the following section.

73

Gleying Few studies have investigated the possible dissolution of iron oxides in wet or waterlogged circum-stances and the extent to which this may affect the magnetic susceptibility of archaeological deposits. Le Borgne (1955) has suggested that maghemite can change into a non-ferrimagnetic iron oxide when gleying occurs in the soil. During the process of gleying Fe(III) is dissolved under long lasting redu-cing conditions and relocated to a higher level, to the zone in which the groundwater fluctuates, causing the typical iron staining. In Broekpolder, the upper part of the soil profile has not been severely flushed from iron, as can be seen in the data from the heating experiment. However, the lowest sample, A80 80-90cm, appears to be affected, probably by gleying. It is likely that this process has added to the concentration of iron in the presumed archaeological layer and the ditch fill. Iron staining, which was observed in the un-disturbed deposits confirms this suspicion. In Harnaschpolder the effect of gleying on the iron mineralogy can clearly be seen by comparing core A180 and A60. Although all the magnetic susceptibilities are low, the amount of iron in the sedi-ments, as can be estimated from fractional conversion measurements, is very different. In core A180, the gleying horizon is present at 45 cm below the surface, below that level the sediment is flushed from iron. Like in Broekpolder, the mobilized iron appears to have accumulated in the layer above, the fractional conversion sample of which acquired a fairly high magnetic susceptibility after ignition. This layer also contains a lot of organic matter. In core A60, the gleyed layer is located at greater depth, starting at 80 cm below the surface. Here, gleying is less likely to have affected the iron mine-ralogy of the archaeological layers, and the fractional conversion data from this core show neither flushing nor enhancement of iron in either of the two layers that were sampled. The depth of the groundwater and gley zone on the location where the samples for IRM measurements were collected is similar to core A60. In the section, iron staining could be observed just above the water table (Fig. 29). Evidence of the gleying process can be seen in the results of the IRM measurements on both sites, as hematite and oxidized magnetite (maghemite) are present in the subsoil, whereas these iron oxides do not usually occur in the temperate climatic zone. Maghemite may however be formed during the oxidation of green rust (Taylor 1980), i.e. during the oxidation of the gley zone. Hematite can form by the dehydration of ferrihydrite, which is a common iron mineral in gleyed horizons. Alternatively hematite and oxidized magnetite may have formed during a heating episode of the soil, but this would not explain the concentration of hematite and oxidized magnetite in the subsoil. In Broekpolder, the presence of these iron oxides in the archaeological deposits probably indicates that this level has been affected by gleying. Leaching The process of leaching occurs, for example, during podzolisation. Iron is leached from the upper part of the soil and redeposited at a lower level, in the case of a podzol in an ‘iron pan’ in the form of (anti-ferromagnetic) lepidocrocite. Coarse (sandy) soils are more easily leached of iron than finer soils or soils with a high organic matter content, and only very acidic conditions can mobilize iron ac-cording to Weston (1999). Detailed groundwater measurements on the Broekpolder archaeological monument (Van Heeringen et al. 2004) have shown that a slight to a more severe acidification of the topsoil layer has taken place. In those locations where a calcareous layer is present, the acidity appears to be neutralized. It is likely that the acidification of the topsoil has caused part of the iron to be flushed to lower layers. This process can account for the observed increased iron content of the archaeological deposits when compared to the topsoil, as manifest in the fractional conversion results. The concentration of dislocated iron from acidification and gleying in the archaeological layer and the fill of the archaeological features in Broekpolder can possibly be attributed to the high cation exchange capacity of the organic matter that is present in these deposits. It is more likely, however, that the iron has precipitated at this level, regardless of its composition. In Harnaschpolder no signs of leaching were observed.

74

Water and seawater logging Weston (1999) has investigated the influence of waterlogging on the iron content and magnetic susceptibility of the soil. He states that waterlogging will influence the magnetic susceptibility of a soil as long as it is severe and long lasting (Weston 1999). Further research is needed to quantify ‘long lasting’ and ‘severe’. Westons research has concentrated on fresh water environments. In marine and estuarine environ-ments however, soils and associated archaeological deposits may be waterlogged with seawater for a considerable length of time. From an archaeological prospection point of view, there has not been any attention to this process of seawater logging, but there has been in earth science. Experimental studies into the dissolution of iron oxides under the influence of sea water show that almost all iron oxides, including those that are critical to the magnetic detectability of archaeological features, can dissolve to form iron sulphides (Raiswell & Canfield 1998, Canfield et al. 1992, Poulton 2004). Experimental data for the time that is needed to dissolve various iron oxides is not conclusive (Table 15), but appears to be of a short duration on an archaeological time scale. Table 15 The reactivity of soil iron oxides to sulphide. After Canfield et al. (1992) and Poulton et al. (2004). iron mineral half-life (Canfield et al. 1992) half-life (Poulton et al. 2004) ferrihydrite 2.8 hours 12.3 hours lepidocrocite < 3 days 10.9 hours goethite 11.5 days 63 days hematite 31 days 182 days magnetite (uncoated) 105 years, (coated) 230 years 72 days A soil that stores iron sulphides in the anaerobic part of the soil section is a Potential Acid Sulphate Soil, and usually occurs in (former) tidal, backswamp or estuarine environments. Dent & Pons (1995) describe how soil iron sulphides are formed. Bacteria can reduce sulphate (SO4) from seawater to H2S, and Fe(III) from the soil to Fe(II) during the decomposition of organic matter. Organic matter is needed as a source of energy for these bacteria, and also for creating anoxic circumstances. This will lead first to the formation of iron monosulphides, like pyrrhotite (FeS) (Berner 1984) and greigite (Fe3S4) (Doner &Lynn 1989), and eventually to pyrite (FeS2). Visual inspection and Curie measurements of the sediments of the anaerobic part of the soil section on the location of the highly magnetic (geological) anomalies that were encountered in Harnaschpolder, suggested that one of the intermediates to pyrite, i.e. greigite, is likely to be causing the unexpected magnetic anomalies in the magnetometer data over the estuarine deposits. High magnetic suscepti-bility layers within these deposits appear to be associated with organic matter, pointing towards the bacterial formation of the iron sulphides. The observed preferential formation of iron sulphides in organic deposits may be of importance for archaeological prospection, as seawater logging may lead to the increased detectability of archaeological features with a high organic matter fill. Draining a Potential Acid Sulphate Soil triggers a series of chemical reactions that eventually transform it into an Actual Acid Sulphate Soil. After the water levels fall sufficiently, oxidation starts and the iron sulphides present in the soil react to form sulphuric acid (SO4) and iron in solution (Fe(II)). In calcareous soils the acidity might be neutralized by calcium carbonate (CaCO3) in the matrix (Ritsema & Groenenberg 1993). Crockford and Willett (1995) carried out an experiment on the oxidation of sulphidic sediments. They found that during oxidation the magnetic susceptibility of the soil material first decreases quickly, then more slowly and eventually rises slightly. This could, according to the authors, be explained by the rapid decomposition of greigite upon exposure to oxygen, followed by the somewhat slower decomposition of pyrite and the formation of a para-magnetic iron oxide, probably ferrihydrite. Research into the type of iron oxide that forms upon the oxidation of pyrite in acid mine drainage has found that goethite can be the end products of this process. In non-calcareous soils the iron can precipitate as jarosite (KFe3(SO4)2(OH)6) creating the yellow mottling that is so characteristic for Actual Acid Sulphate Soils. It is inevitable that the iron mineralogy of a soil changes during seawater logging. Further experi-mental data are needed to identify the transformations of soil iron that can occur.

75

For Broekpolder and Harnaschpolder the following model is proposed: (1) before the sea / brackish water inundation, topsoil material and the fill of archaeological features

has possibly had an iron composition similar to the modern day topsoil; (2) during the inundation, most of the iron is brought into solution by iron reducing bacteria during

persistent seawater logging; (3) after the inundation, the iron precipitates during oxidation, either as ferrihydrite, or as goethite. Apart from sea or brackish water logging, it is possible that the level of the superficial ground water, which in Broekpolder is currently sweet water (Van Heeringen et al. 2004), has fluctuated in the past. Reduction of iron oxides under the influence of sweet water is a slow process (Weston 1999) but the subsequent oxidation of the sediment would follow the same process as with seawater logging, with goethite as an outcome. Such changes in the past may have left very little evidence in the magnetic soil record but the magnetically homogenised nature of the deposits is compatible with such expla-nation. 5.4.6 Conclusions In the proposed scenario, the iron mineralogy of the investigated sections in Broekpolder and Harnaschpolder has changed considerably due to post depositional processes like gleying, leaching and (sea)water logging. It is likely that the iron composition of the undisturbed matrix and the archaeological deposits has become homogenized due to prolonged water or seawater logging. During sea or brackish water logging of the soil, iron is brought into solution, and iron sulphides may be formed. The highly magnetic anomalies that were encountered in the magnetometer survey of Harnaschpolder are likely to be caused by concentrations of iron sulphides, most noticeably greigite, in the anaerobic part of the soil. Although the strong magnetic anomalies hamper magnetometer surveys for archaeological purposes, the preferential formation of iron sulphides in organically rich deposits may enhance the detectability of archaeological features in waterlogged conditions. After the inundation and after the lowering of the groundwater table, gleying has caused iron to move up in the soil profile, whereas in Broekpolder the acidification of the topsoil has caused iron to move down in the soil profile. In Broekpolder, the iron of either or both of these processes has precipitated at the level of the presumed archaeological layer and the archaeological features, resulting in a low magnetic susceptibility and an undifferentiated iron mineralogy between the archaeological deposits and the undisturbed matrix. In Harnaschpolder, gleying has had less effect on the archaeological layers in the eastern part of the site, and the iron mineralogy of the archaeological deposits differs from the undisturbed matrix. The oxidized magnetite / maghemite and hematite, which is occurring in the subsoil at both sites, and in the archaeological deposits in Broekpolder, is thought to be the result of the gleying process. The fill of the modern feature at Broekpolder, has not (yet) been exposed to the processes that have been described above, its iron mineralogy and magnetic susceptibility are similar to the topsoil iron mineralogy and susceptibility. The modern feature could be mapped in the magnetometer survey that was conducted on the archaeological site, in contrast to the archaeological features, which could not be mapped, probably because of changes in their iron mineralogy as a result of the proposed post depositional processes.

76

77

6 Wind blown and fluvial deposits 6.1 Wind blown sands The eastern and southeastern parts of The Netherlands are characterized by the superficial occurrence of Pleistocene aeolian, fluvial and glacial deposits. This section is concerned with the wind-blown sands, which include Late Pleistocene cover sands and Holocene inland dunes. The cover sands constitute approximately 40% of the superficial deposits and are found predominantly on the higher grounds. They were formed in the Weichselian Late Pleniglacial and Late Glacial and have been dated between circa 16,000 and 11,500 BP. Based on the relief and the underlying deposits, the cover sand area can be divided into the north, the central, the eastern and southern cover sand area. The cover sands of the northern sandy area overlie deposits from the Saalien ice age, extensive boulder clay deposits, which have been pushed up by the land ice in the southern part of the area. The Weichselien cover sand relief follows the Saalien relief, which makes it more pronounced than in the southern cover sand area. A rising ground water table has facilitated the peat formation in the Holocene, starting in the lower parts of the landscapes and in the river valleys, and continuing to include the higher parts. The excavation of peat as a fuel started around 1600, causing the Weichselien cover sand to surface once again in many places. The northern sandy area has not been investigated in this study because of the relative lack of intrusive archaeological investigations taking place, which were needed for the ground truthing of the magnetic data. The central sandy area, of which the site of Den Dolder is an example, is dominated by high ice pushed ridges from the Saalien ice age. The material of these terminal moraines consists of reworked coarse river deposits. The Weichselien cover sands occur in the lower parts of the landscape and on the slopes of these ice pushed ridges. Extensive areas of drifting sand with inland dunes have formed in those areas that lacked sufficient vegetation. A similar situation occurs in the eastern cover sand area, but here the ice pushed ridges from the Saalien ice age are lower than in the central area. In this region the sites of Heeten and Raalte have been investigated for this study. The southern cover sand area is the only sandy area that has not been covered by land ice during the glacial periods. The resulting relief of the cover sands is minimal, and consists of low cover sand ridges and shallow valleys. As a whole, the southern Pleistocene sand area dips from the southeast to the northwest. East of the river Meuse, the cover sands are deposited on top of the Pleistocene river terraces. The area is drained by rivers that flow through the valleys that were shaped earlier, during the glacial period. The southern cover sand area is represented by two sites, Breda and Slabroek. Inland dunes are derived from the finer fraction of the Weichselien cover sands, which makes these blown out deposits younger than the parent material that they originate from. The cover sands, and the inland dune material, is non calcareous. Coastal dunes have a fine texture and are calcareous south of the breach in the dune row at Bergen, and non calcareous to the north of it. During this study, no archaeological sites have been investigated on inland dunes, whereas coastal dunes are represented by the site of Limmen. Soil formation on sandy soils can lead to the development of podzols. The type of podzol that has been formed depends mainly on the hydrological situation, and on topography. Starting in the Prehistory, human impact changed the landscape. On the higher grounds the fertility of the sandy soils was insufficient to grow crops, but the topsoil could be excavated to be used elsewhere as plaggen. These plaggen were added to the medium high and lower sandy soils. This way, fertile plaggen soils were constructed, a process that probably started in the Iron Age.

78

The excavation of plaggen contributed to the instability of the higher grounds by making the higher parts of the landscape prone to wind erosion. This erosion is one of the causes of the formation of inland dune areas. Because of the age of the cover sands, most of the archaeological remains in the sandy areas can be expected on or just underneath the current land surface. Only Paleolithic archaeological sites are stratigraphically located underneath the cover sands. Moreover, Mesolithic and Neolithic archaeo-logical remains are more likely to be covered by redeposited cover sand material (e.g. drifting sands) than younger archaeological remains. The sites that are included in this study are all of a younger age. 6.1.1 Magnetic susceptibility of the cover sands Soil samples for magnetic susceptibility measurements were taken at the sites of Raalte, Heeten, Breda and Den Dolder (Fig 31). The samples were initially collected to investigate magnetic suscepti-bility contrasts on the individual archaeological sites. They do also give an insight, however, to the range of magnetic susceptibility values that may be expected on wind blown sands. The mean values for the soil horizons that were encountered at these sites and for the fill of the archaeological features that were sampled are displayed in Figure 31. Details about the individual sites can be found in Appendix I. For magnetometer surveys to successfully detect archaeological features, the fill of these features needs to have a different, usually enhanced magnetic susceptibility (see Chapter 3). A high magnetic susceptibility topsoil is an indication of the potential magnetic susceptibility enhancement of the soil.

0 10 20 30 40 50 60 70 80

C-horizon

BC-horizon

B-horizon

archaeological feature

E-horizon

topsoil RaalteHeetenDen DolderBreda

C-horizon BC-horizon B-horizon archaeological

feature E-horizon topsoil

Raalte 10.53 (N=2) - - - - 8.53 (N=1) Heeten 9.6 (N=7) 67.25 (N=4) 27.99 (N=7) 14.48 (N=7) 22.43 (N=21) Den Dolder 0.94 (N=2) 2.42 (N=2) 2.4 (N=2)1 2.77 (N=16) 7.48 (N=2)2 7.7 (N=3) Breda 1.51 (N=4) - - 0.99 (N=4) - 12.85 (N=4) 1 only the samples from the wind blown sand have been included; 2 horizon has been partly mixed with topsoil material by ploughing. Figure 31 A comparison of the magnetic susceptibility through the soil profile on wind blown sand sites. The value that is displayed is the mean value of the magnetic susceptibility of a number of N samples.

79

The topsoil samples from Breda and Den Dolder are higher in magnetic susceptibility than the subsoil samples. If features are (partly) filled with topsoil material, and the magnetic susceptibility of the soil profile decreases with depth, the archaeological features can cause a magnetic anomaly. In the soil profile of Den Dolder a magnetic susceptibility decrease with depth can be seen, but the enhancement of magnetic susceptibility of the archaeological feature fills is minimal. The samples from Breda show that a magnetic susceptibility contrast exists between the topsoil - a plaggen soil - and the subsoil, but the archaeological feature fills are not enhanced. For Raalte and Heeten, the magnetic susceptibility of the soil layers does not appear to decrease with depth. At the Raalte site very few samples have been collected, but the magnetic susceptibility of the subsoil is greater than the topsoil susceptibility. In Heeten, which was sampled much more intensi-vely, both by hand auger and from excavation trenches, the magnetic susceptibility of the B- and the BC-horizon surpasses the topsoil susceptibility, whereas the archaeological features actually have a lower magnetic susceptibility than the topsoil. As a result, the magnetic susceptibility contrast between the archaeological features and the B- and the BC-horizon which they are cut into is negative. The contrast between the fill of the archaeological features and the undisturbed subsoil, however, will be positive. The complexity of this situation is illustrated in Figure 32. There are both positive and negative magnetic contrasts within the same archaeological feature. The shape, size and sign of the resulting magnetic anomaly that is caused by the feature depends on the amount of contrast and the volume of each of the components, and it is not possible to predict from magnetic suscepti-bility samples only if these contrasts will cause a detectable magnetic anomaly.

Figure 32 An example of the change of sign in magnetic contrasts for features with a homogeneous fill in a magnetically layered subsoil. The presence of high magnetic susceptibility subsoil layers can cause problems for the magnetic detection of archaeological features for two reasons, because of the sign reversal of the contrast, as was discussed above, and because of the variability of magnetic susceptibility, which will be dis-cussed below (§ 6.1.5). At this point it is interesting to look into possible causes for the observed lack of magnetic susceptibility enhancement in archaeological feature fills on sandy soil and for the presence of high magnetic susceptibility subsoil layers in Raalte and Heeten. Why do the archaeological features on sandy soil lack a magnetic susceptibility enhancement? Difficulties with magnetic prospection on coarse mineral sand have been encountered and investigated by Weston (Weston 1999). Because of the lack of any detectable magnetic contrast that he encountered on the archaeological site of Easingwold, United Kingdom, where in subsequent excavations indications of intense heating on the site were encountered. Laboratory experiments were conducted in order to explain why these high temperature features, or indeed any of the features, could not be detected magnetically. It was found that texture played an important role in the acquisition and the conservation of magnetic susceptibility enhancement (Weston 2004). Sandy soils achieve their maximum magnetic susceptibility at higher temperatures than finer textured soils. There are two factors that suppress the level of enhancement, staged heating, particularly in sandy soils, and heating under waterlogged conditions in mineral soils.

80

Heating to lower temperatures will cause any organic matter present to combust, any goethite (the main iron oxide in under temperate humid conditions) will convert into hematite, but a lack of further organic material will prevent the reducing circumstances that would be necessary for the further con-version into magnetite / maghemite. Experiments on leaching showed that sandy soils are more prone to the loss of iron than fine textured soils, except when they have been ignited, in that case the Fe present in the matrix is more resistant to leaching. Why do high magnetic susceptibility subsoil layers occur in Raalte and Heeten but not in Breda and Den Dolder? There is a marked difference between the magnetic susceptibilities from Den Dolder and Breda on the one hand, and Raalte and Heeten on the other. These differences may be caused by the site function, the former two being settlement, burial, and off-site sites, and the latter two being iron production sites. It is possible that a magnetic susceptibility enhancement of the undisturbed soil layers has occurred because of the repeated high temperature activities that have taken place on the soil surface. An indication that this process has taken place is the magnetic enhancement of the B- and BC-hori-zon. There horizons contain much iron, but as was seen in Chapter 3, the iron oxides usually occur as ferrihydrites, with a low magnetic susceptibility. Transformation through dehydroxylation can change ferrihydryte into hematite or via goethite into maghemite, resulting in a higher magnetic suscepti-bility. 6.1.2 Magnetic anomalies in the cover sands Very few magnetic anomalies that were mapped on the sandy soil sites can be attributed to archaeological features. In Breda and Slabroek, no archaeologically relevant interpretation could be made of the magnetometer data (see Appendix I, 6 and 20). In Heeten, a late Roman Period metal production site, two types of archaeological features had a magnetic response; furnaces and large rubbish pits. In Den Dolder and Raalte no magnetometer surveys were carried out. The results of the magnetometer surveys in Heeten and a comparison with the excavation data are displayed in Figure 33. The furnaces, marked with a star, cause strong, irregular magnetic anomalies. The positive and negative component of the magnetic anomaly are oriented randomly, which is an indication that the anomaly is caused by remanent magnetism. The shape and size of this type of anomaly is independent of the type of geology that is surveyed. Further details about the furnaces can be found in § 7.4. The anomalies that are caused by rubbish pits are shown with arrows in the magnetic plot. They consist of a positive anomaly with a small negative component. It is likely that these anomalies are caused by induced magnetization in features with a magnetic susceptibility contrast between the fill of the feature and the matrix. It is remarkable that the furnaces and these two pits are the only apparent archaeological features in the magnetometer data. As was discussed above, the fills of the archaeological features that were sampled for magnetic susceptibility measurements had a lower magnetic susceptibility than the matrix they were embedded in, but there are no negative anomalies that reflect the presence of archaeological features in the data. The anomalies may have been 'cancelled out' because of differences in magnetic susceptibility in the subsoil, as was illustrated in Figure 32. If the non-archaeological soil is assumed to be a homogeneous medium, however, whether or not a feature can be detected in a magnetometer survey depends on the strength of the magnetic contrast, the volume of the archaeological feature and the depth of burial. A combination of these three variables may explain lack of detectable magnetic anomalies in terms of strength and size, but the visual detection of the anomalies may also have been hampered by the variability of the magnetic susceptibility within the soil layers, either in the topsoil or subsoil. In the following paragraphs both masking (increase of depth) and variability are discussed. 6.1.3 Masking Masking takes place when the archaeological record is covered by a later deposition of soil material. In the sandy area, there are two main types of masking material, plaggen and wind blown sand deposits (inland dunes).

81

If the post archaeological deposit is assumed to be a homogeneous layer, the only influence of the addition of such a layer on a magnetometer survey is that the distance between the archaeological record and the surface is increased. Any magnetic anomalies are further away from the magnetometer, resulting in a weaker anomaly strength (see § 3.9.2). Figure 34 shows an example of the decrease of the anomaly strength with distance to the magnetometer.

Figure 33 The results of the magnetometer survey (top) and the excavations (bottom) in Heeten. Furnaces that were detected magnetically are indicated with a star, two rubbish pits that have caused a magnetic anomaly with an arrow.

82

0

5

10

15

20

25

0.75 1 1.25 1.5 1.75 2

distance between top of the body causing the anomaly and the magnetometer (anomaly: 0.5 x 0.5 x 0.5 m; magnetic susceptibility contrast = 185)

peak

val

ue m

agne

tic re

spon

se (n

T)

Figure 34 The peak value strength of the magnetic anomaly decreases when the distance between the object and the magnetometer is increased. 6.1.4 Variability The deposits masking the archaeological layers are rarely magnetically homogeneous. The material that makes up a plaggen soil, for example, has often been mixed with household waste. Inclusions like brick and small metal objects cause remanent magnetic variations that do not reflect the archaeo-logical record. Moreover, variations in the magnetic susceptibility within the post archaeological layer can cause induced magnetic anomalies that may resemble archaeologically caused anomalies. If this remanent or induced magnetic variation is expressed at the surface in a similar way as the magnetic variation that is caused by an archaeological feature, an archaeological magnetic anomaly may be present, but it may not be recognized as such. The variation in magnetic susceptibility values that were measured in Heeten are displayed in Figure 35.

0.00

20.00

40.00

60.00

80.00

100.00

120.00

HH03topsoil

HH04topsoil

HH03feature?

HH04posthole

HH03 B HH03 BC HH03 C

magnetic susceptibility x 10-8 m3/kg

maximum

minumum

median

Figure 35 Minimum, maximum and median magnetic susceptibility for the topsoil, B-, BC- and C-horizon and that archaeological features that were sampled. HH03 samples were collected in the 2003 season by hand auger, HH04 samples have been taken directly from the excavation in 2004.

83

It is clear from the data that in Heeten there is a contrast between for example the median magnetic susceptibility of the archaeological features and the B- and BC-horizon, but the variations within the B- and BC-horizon are much larger. The variation in these subsoil layers may cause non-archaeo-logical magnetic anomalies. The magnetic susceptibility variation in the plaggen soil (topsoil), how-ever, is smaller, but may still cause anomalies because the magnetometer reading is more influenced by the topsoil layer than the archaeological layer. Any possible magnetic susceptibility variation in the C-horizon in Heeten is not reflected by the samples that are displayed in Figure 35, that were taken by hand auger before the excavation. During the excavation in Raalte another possible cause of magnetic variation was encountered as the C-horizon appeared to have obtained an orange-red colour in places (Fig. 36), the colouring occurs in patches. During the excavation in Heeten the red sand was encountered here as well. This hydro-logical phenomena is known to occur on sandy soils in the central and eastern sandy area; Bakker & Rogaar (1993) have explained it to be caused by the local abundance of hematite and magnetite in the soil matrix. The bright orange-red colour of the sand supports the explanation of the presence of hematite. The authors suggest that these iron minerals have been transported through effluent seepage.

Figure 36 Excavation trench in Raalte. Patches of red sand (grey in picture) can be seen in the foreground. Darker patches in the background are archaeological features. The results of initial, limited magnetic susceptibility measurements that were carried out within the framework of this study, suggested that the red sand at Raalte had a higher magnetic susceptibility than the 'non coloured', yellow sand (Appendix I, 19). A more detailed investigation into the magnetic susceptibility of the patches of red sand could be conducted at Heeten Hordelman, where similar red sand patches had come to light during the excavations. An area that included red sand, non coloured sand and archaeological features was sampled for magnetic susceptibility measurements. The samples were taken in a one meter grid from the bottom of the excavation trench (top of the C-horizon). The results of the measurements have been displayed as a plan in Figure 37. In the western half of the survey, the archaeological features have enhanced magnetic susceptibilities, in the eastern half the red sand area can also be defined by higher magnetic susceptibility values. The magnetic susceptibility in the fills of the archaeological features in this example is higher than the red sand for two of the archaeological features, similar for two others and lower for the fifth feature. It has to be kept in mind that the plot only represents the superficial magnetic susceptibility values of the top of the C-horizon.

84

Figure 37 Magnetic susceptibility survey in an excavation trench in Heeten. See the plan in Figure 33 for the location of this survey. Soil samples were collected at the grid intersections and measured for magnetic suscepti-bility in the laboratory. Five archaeological features are included in the western half of the grid, the red sand patch is located in the eastern half. Both the archaeological features and the red sand have an enhanced magnetic susceptibility. The strength of the magnetic anomalies that may be caused depends not only on the magnetic susceptibility contrast, but also on the volume of the bodies that cause them. It is clear to see, however, that the range of magnetic susceptibility values of the archaeological feature fills and the red sand is similar. This may cause problems with false positives, and it may also make magnetic anomalies that are caused by induced magnetization of archaeological features less easy to define. 6.1.5 Magnetic susceptibility of coastal dunes At the excavation of the Medieval settlement of Limmen, samples could be collected from natural and archaeological contexts in coastal dune sands. It can be observed that there is a clear differentiation in magnetic susceptibility between the topsoil, the subsoil and archaeological features (Table 16). The magnetic susceptibility of the undisturbed matrix is low, and comparable to the susceptibility of marine clastic sediments. The archaeological features generally have a positive magnetic contrast compared to the matrix that they are embedded in and they can be expected to cause positive magnetic anomalies. The magnetic susceptibility of the topsoil is higher than the subsoil susceptibility and varies in the same range as the values for the archaeological features, which could cause problems with masking and variability (see § 6.1.3 and § 6.1.4). Table 16 The results of the magnetic susceptibility measurements on the samples taken from the excavation at Limmen. All values x 10-8 m3/kg. N is the number of samples. For details see Appendix I, 12.

maximum minumum median mean N topsoil 18.90 11.43 14.72 15.15 7 archaeological feature 84.13 4.84 13.39 17.15 42 undisturbed 8.15 4.68 6.06 6.31 9

85

One set of features proved to have a much higher magnetic susceptibility, in the fill of two ditches that defined the course of the Medieval road, values over 200 x 10-8 m3/kg were measured. The reason for this enhancement is unknown.

Figure 38 The results of the magnetometer survey at Limmen (top) and the interpretation diagram in black (bottom) as an overlay over the results of the excavation (grey). Magnetic anomalies that have been caused by metal objects have been hatched in the interpretation diagram.

86

6.1.6 Magnetic anomalies in coastal dunes The difference in response between the generally slightly enhanced archaeological features and the strongly enhanced road side ditches in the magnetometer survey can be seen in Figure 38 and 39. The ditches show clearly as parallel linear anomalies, whereas most of the other archaeological features do not cause a detectable magnetic anomaly. Most of the anomalies that were mapped, on the other hand, represent archaeological features, almost all of which are wells. It is not clear which part of the well or well fill causes the magnetic anomaly. Most of the wells that were detected are constructed with plaggen, rich in organic material, the presence of which is a prerequisite for the enhancement of magnetic susceptibility. The primary fill of the wells is usually also high in organic material content, whereas the secondary fill usually is not, samples from the upper fill of one of the wells confirmed that these non organic deposits had a low magnetic susceptibility (see Appendix I, 12 for details). It is likely that the anomalies that are caused by the wells must be attributed to a combination of the high magnetic susceptibility construction material or primary fill and the large volume of the high mag-netic susceptibility body, if compared to for example a pit or a ditch on the same settlement site. General conclusions about the possibilities for magnetic prospection on coastal dunes can not be made as only one archaeological site has been surveyed on this geological background. For this one site, however, a general pattern of magnetic enhancement of the fill of archaeological features could be seen. Some of these features caused anomalies that were detectable with the instrument that was used. The use of a more sensitive instrument would probably have mapped more anomalies that were related to the archaeological record.

Figure 39 A summary of the non-remanent magnetic anomalies that were identified and their relation to the archaeological record. Only the northern (excavated) half of the interpretation diagram in Figure 38 is shown. Anomalies that did not prove to be caused by an archaeological feature: grey line (no fill); early period (800-1000 AD): black line (no fill); middle period (1000-1150 AD): light grey fill; late period (1150-1250 AD): dark grey fill; post settlement: black fill. See Appendix I, 12 for details. 6.2 Loess Like the cover sands, the superficial deposits of loess in the southeast of the country have been deposited around 10.000 BC, the late-Weichsel period. They cover an older (Saalien) loess deposit. The Dutch loess area lies at the northwestern edge of the much larger Belgian-Germanic loess region. The relief in the Dutch loess deposits is mainly determined by the underlying terraces of the river Meuse and by the tectonic breaks in the area. Only one archaeological site has been investigated on loess in this study, a Roman villa in Meerssen.

87

6.2.1 Magnetic susceptibility The limited amount of work that has been done on loess shows promising results. A selection of the results of the measurement of magnetic susceptibility samples that were collected during the exca-vation of the Roman villa at Meersen is shown in Figure 40. Although limited in number, the samples show a clear magnetic contrast between the undisturbed matrix and the fill of the archaeo-logical features. A similar contrast can be seen between the undisturbed matrix and the samples that were collected from the fill of a modern ditch, these samples have magnetic susceptibilities of 29.2 and 30.67 x 10-8 m3/kg (not displayed in Fig. 40).

0

5

10

15

20

25

30

35

40

45

topsoil feature fill undisturbedloess

mag

netic

sus

cept

ibili

ty x

10-8

m3 /

kg

maximum

minimum

median

Figure 40 The minimum, maximum and medial magnetic susceptibility of samples from Meerssen. Note the limited number of samples: topsoil N = 3, feature fill N = 4, undisturbed loess N = 3. 6.2.2 Magnetic anomalies The magnetic contrast that was observed in the samples is a prerequisite for the formation of induced magnetic anomalies. Nevertheless, when the interpretation of the magnetometer survey at Meerssen is compared to the results of the trial trenches (Fig. 41), only very few of the archaeological features have given a magnetic response. These include remanent magnetic features and two anomalies that are likely to be caused by buried limestone walls, anomalies which are not caused by magnetic susceptibility enhancement. Two pits and a modern ditch are represented by induced magnetic anomalies. It is interesting to speculate why these features do and the remaining features do not cause detectable magnetic anomalies. The fill of the modern ditch has a magnetic susceptibility comparable to the feature fills that were sampled in trench 1 (the western trench), the ditch has a greater volume and causes an anomaly, whereas the much smaller pits do not. During the trial trenching it was observed that the top level of the archaeological remains had been destroyed, for example, the foundation trenches were the only remnants of the Roman building. The shape and size of a magnetic anomaly is influenced by the amount of magnetic contrast and the volume of the body causing the anomaly. The way the anomaly is recorded also depends on the distance between the body and the magnetometer. Based on the limited evidence that was collected, there appears to be a magnetic contrast between the fill of the features and the loess matrix that they are embedded in. It is likely that magnetic anomalies have been caused by the archaeological features, but most of these could not be mapped at the surface in the magnetometer survey. The combination of the amount of magnetic contrast, the volume of the features and the depth of burial has been detrimental to the detection of the anomalies at surface level. Magnetic mapping at less damaged sites on loess geology would probably be more successful. It is possible that a more sensitive magnetometer would have mapped a greater amount of archaeologically relevant anomalies.

88

Figure 41 The interpretation of the results of the magnetometer survey (in grey) combined with the trial trench results (in black) at the Meerssen Roman villa site. For more details of the survey and the magnetometer data refer to Appendix I, 13. 6.3 Fluvial deposits The extent of the river area in The Netherlands is defined by the presence of Pleistocene and Holocene deposits of the rivers Rhine and the Meuse. It consists of a central river area, west of the line Arnhem-Nijmegen, the valley of the river IJssel, and the valley of the river Meuse in the southern part of the country. West of the line Utrecht-Den Bosch lies the perimarine part of the river area. In the central river area Holocene deposits overlay Pleistocene deposits. Geomorphologically the area is dominated by natural levees of a calcareous nature, with coarse material in the subsoil, and by backswamps, which, because of their slow sedimentation rate, consist of material of a fine soil fraction, clay and silt. Four archaeological sites were investigated in the central river area in the course of this study; Deil, Meteren, Wijk bij Duurstede and Zaltbommel. The soil material on these sites is calcareous clayey silt and silty clay. In the IJssel valley the Holocene deposits are much thinner than in the central river area. The IJssel valley has not been represented by any archaeological sites in this research. Neither has the perimarine area, in this westernmost part of the river area. Here, the sedimentation of the rivers and peat formation has depended largely on sea level changes. In the Meuse valley the Holocene deposits are concentrated in the river valley, whereas the higher terraces that surround the valley have been deposited in the Pleistocene era. The Meuse deposits are and have been non-calcareous. Two archaeological sites were investigated in the Meuse valley; Beugen en Borgharen. In a previous study (Anderson, not published) a magnetometer survey was conducted in Gennep, also in the Meuse valley.

89

6.3.1 Magnetic susceptibility In Figure 42 the mean magnetic susceptibility of the soil samples that were collected on sites on fluvial geology is compared. The first observation that can made is that the values for the undisturbed subsoil are all low, and topsoil values are much higher. A good magnetic susceptibility contrast between topsoil and subsoil can be an indication of the suitability of a site for a magnetometer survey, because it is an indication that the sediment can be magnetically enhanced. Moreover, if a topsoil with an enhanced magnetic susceptibility is present now, it may also have existed during past habitation. Negative archaeological features can be (partly) filled with topsoil material, the topsoil-subsoil contrast is extended to the feature fill-subsoil contrast, a contrast which may result in induced magnetic anomalies under the influence of the earth’s magnetic field. The topsoil-subsoil contrast gives an indication of a possible magnetic contrast between the fill of archaeological features and the subsoil. Looking at the magnetic susceptibility of the samples from archaeological features it is clear that this contrast is indeed present in three out of the five sites that have been investigated on fluvial geology in this study. In Borgharen and Beugen the magnetic susceptibility of the feature fills is actually higher than the topsoil values. This is illustrated in Figure 43, which displays a section through an archaeological layer or feature in Beugen. One of the features in Borgharen (feature 253 in trench 11, see Appendix I, 7 for details) has a magnetic contrast with the matrix surrounding it of 182-15 = 167 x 10-8 m3/kg. A pit that measures 0.5 x 0.5 x 0.5 meter and is buried at a depth of 0.5 meter would give a magnetic anomaly with a peak of 12.4 nT, which would be easily recognized in the results of the fluxgate gradiometer that was used during this project. Based on the magnetic susceptibility measurements, both of the sites that have been investigated in the Meuse valley have a good potential for a magnetometer survey.

0.00 50.00 100.00 150.00 200.00 250.00

undisturbed

archaeological layer

archaeological feature

topsoil

Borgharen

Beugen

Zaltbommel

Wijk bij Duurstede

Deil

undisturbed archaeological layer archaeological feature topsoil central river area Deil 4.7 (N=3) - 4.6 (N=2) 9.6 (N=3) Wijk bij Duurstede 11.0 (N=2) - 75.0 (N=3) 88.7 (N=3) Zaltbommel 8.0 (N=5) - 7.3 (N=12)1 20.3 (N=5) Meuse valley Beugen2 4.44 (N=12) 16.74 (N=8) 48.58 (N=9) 30.64 (N=12) Borgharen 10.5 (N=6) - 195.2 (N=3)3 40.0 (N=4) 1 the anomalous sample with a magnetic susceptibility of 119.5 x 10-8 m3/kg has been excluded; 2 samples from the prehistoric site only; 3 samples from high temperature features have been excluded. Figure 42 A comparison of the magnetic susceptibility of the subsoil, archaeological layer and archaeological feature fills and the topsoil on fluvial sites. The value that is displayed is the mean value of the magnetic sus-ceptibility x 10-8 m3/kg of a number of N samples.

90

Figure 43 The downcore magnetic susceptibility of feature X1 of the Beugen-Zuid survey. Samples have been taken by hand auger (see Appendix III for full data). High magnetic susceptibility values are associated with the undisturbed archaeological level, the archaeological deposits that were plough damaged have medium high magnetic susceptibility values. In Wijk bij Duurstede the samples show a good contrast between the fill of archaeological features and the matrix that they are embedded in, but topsoil samples have an even higher magnetic sus-ceptibility. Based on the magnetic susceptibility contrast this site may be suitable for a magnetometer survey. In Zaltbommel and Deil no such contrast could be observed in the samples that were collected here, which makes it less likely that these sites in the Central river area can be mapped magnetically. 6.3.2 Magnetic anomalies For the Meuse valley sites of Beugen en Borgharen, the contrast that could already be observed in the magnetic susceptibility samples (§ 6.3.1) is reflected in the results of the magnetometer surveys on these sites. In Figure 44 the results of the survey in Beugen and in Figure 45 of the survey in Borg-haren are displayed. On both archaeological sites, part of the archaeological features present in the subsoil caused a magnetic anomaly that could be detected in the fluxgate gradiometer survey. In Beugen, pits and ditches could be magnetically detected. Looking at the feature density in the areas that were excavated around the magnetometer survey, it is unlikely that all archaeological features that are present have been magnetically mapped. Further intrusive investigations can possibly clarify why certain features do and others do not cause a detectable magnetic anomaly. In Borgharen, archaeological features from different periods are interpreted to have caused a magnetic anomaly, building remains from the Roman Period, early Medieval graves and a number of pits of undetermined age. Another interesting aspect of this survey is the clear plough marks in the magnetometer data, these are an indication of the contrast that exists between the topsoil and the undisturbed subsoil. A dataset that has not been displayed was collected by Anderson in Gennep (Anderson, not published). A small Fluxgate gradiometer (FM18) survey of 20 x 40 meter was conducted over part of a 4th/5th century settlement on sandy Meuse deposits. The results of the magnetometer survey showed a number of linear and non-linear anomalies, which could - in the excavation after the survey- be interpreted as three ditches, two rubbish pits and a furnace. There were two false positives in the data, but an overall good correlation between the archaeological record that was excavated and the results of the magnetometer survey.

91

Figure 44 The results of the magnetometer survey in Beugen with the interpretation of the magnetic anomalies. For details see Appendix I, 2. In the Meuse valley, the contrast that was observed in the magnetic susceptibility samples translated into detectable magnetic anomalies in the results of the magnetometer survey. Based on the lack of contrast in samples from two of the Central river area sites, Deil and Zaltbommel, it was expected that a magnetometer survey conducted here would be less successful in mapping archaeological features. In Zaltbommel none of the archaeological features that were excavated after the magnetometer survey could be mapped. A second magnetometer survey in the Central river area was conducted on the archaeological site of Meteren, prior to its excavation. On this Iron Age / Roman period settlement site there was a relatively large variation in the recorded magnetic dataset (see Appendix I, 14). Most of the anomalies could be attributed to post-Medieval ditches, although there is some indication that the location of the houses that were excavated show in the magnetometer data as positive anomalies, possibly caused by patches of enhanced magnetic susceptibility soil. 6.4 Conclusions Cover sands, general

• Subsoil magnetic susceptibilities are generally low. Topsoil deposits (plaggen soil) have a higher magnetic susceptibility, but the plaggen soil is not the original topsoil. The fill of archaeological features is not enhanced and has little magnetic contrast to the surrounding matrix.

• Magnetometer surveys do not show any archaeologically relevant induced magnetic anomalies. • Problems with the magnetic detection of archaeological features on coarse mineral soils have also

been reported elsewhere, and are confirmed by the results in this study.

92

• The presence of a plaggen soil increases the distance between the archaeological features and the magnetometer (masking), which results in a weaker signal.

• Magnetic susceptibility variations in red sand and in plaggen soils and remanent magnetic inclu-sions in plaggen soils can cause false positives and can hamper the interpretation of magnetometer data.

Figure 45 The interpretation diagram of the magnetometer survey in Borgharen; see Appendix I, 7 for the data and more details. Cover sands, metal working

• On the metal working site of Heeten, high magnetic susceptibility subsoil layers are encountered, which are probably related to the high temperature activities on the site. Variations in these layers can cause false positives and hamper the interpretation of magnetometer data. Archaeological feature fills have a negative magnetic contrast to the surrounding matrix.

• Magnetometer surveys could not map these negative contrast features A number of positive anomalies, however, all of the furnaces and a number of pits, could be mapped.

93

Coastal dunes • Topsoil magnetic susceptibility is enhanced and there is a clear magnetic contrast between the fill

of archaeological features and the undisturbed matrix. • A multitude of archaeological features could be mapped in the magnetometer survey, with only a

few false positives.

Loess • Topsoil magnetic susceptibility is enhanced and there is a clear magnetic contrast between the fill

of archaeological features and the undisturbed matrix. • A magnetometer survey on a Roman villa site mapped a number of remanent and induced magne-

tic anomalies that were related to the archaeological record. A greater number of features, how-ever, could not be mapped, which was probably due to the damaged nature of the site.

Fluvial deposits, Central river area

• Topsoil magnetic susceptibility is enhanced, but there is no magnetic contrast between the fill of the archaeological features and the undisturbed subsoil on two of the three sites.

• Magnetometer surveys generally did not reflect the archaeological record, although some post-Medieval features could be mapped.

Fluvial deposits, Meuse valley • Topsoil magnetic susceptibility is enhanced and there is a clear magnetic contrast between the fill

of archaeological features and the undisturbed matrix. • Part of the archaeological features could be mapped in a magnetometer survey.

94

95

7 Magnetic mapping of archaeological features in The Netherlands In this chapter examples are given of the magnetic anomalies that are caused by archaeological features in The Netherlands. For each type of archaeological feature, e.g. pit, ditch and well, one or more sets of magnetometer data, which have been collected within the framework of this study, have been displayed. Most of the interpretations have been confirmed by excavation, hand augering or evidence from historical maps. In some cases, the magnetometer data has been interpolated or filtered to improve the clarity of the graphical representation of the data, this has been mentioned in the caption of the figures. All the data that is presented here are subsets of larger magnetometer datasets, the data can be viewed in its original context in Appendix I, where additional information on the sur-veys can also be found. 7.1 Settlements 7.1.1 Pits Continued on next page with Figure 46.

96

Figure 46 Examples of the magnetic response of buried archaeological pits. Pits generally cause a positive mag-netic anomaly with a negative halo or a negative trough north of the positive component. a: A rubbish pit (right hand side) is represented by an oval shaped positive magnetic anomaly (induced) with a negative halo. In the center the response of a remanent object or feature can be seen (e.g. a piece of metal), in this case the core of the anomaly is negative, and the halo around it is negative, the opposite of an induced magnetic anomaly. Data from Heeten, confirmed by excavation. Data range -3 to 3 nT. b: Series of seven possible pits in Meerssen, each pit is defined by a positive magnetic anomaly and a negative trough on its north side. The two southern pits have been confirmed by excavation. Data range -2 to 4 nT; data has been interpolated. c: Two possible pits in Beugen, which are mainly defined by a positive magnetic anomaly. The interpretation has not yet been confirmed by excavation. Data range -2 to 3 nT; data has been interpolated. d: Three possible pits in Borgharen. Note the difference between the pits - positive magnetic anomalies with a small negative component, typical induced anomalies- and the magnetic noise around the proposed pits - small positive anomalies with a larger negative component or with a randomly oriented negative trough, typical responses of remanent objects like bricks or small pieces of metal. The interpretation has not yet been confirmed by excavation. Data range -3 to 3 nT; data has been interpolated.

97

7.1.2 Ditches

Figure 47 Examples of the magnetic response of buried ditches. a: Four ditches in Limmen, which have been confirmed by excavation. At the bottom of the figure a patchy positive linear runs SWW-NEE, the second ditch is located in the middle of the figure and runs in approximately the same direction. The third ditch, in the northeast corner, produces a far more consistent anomaly and proved to have a high magnetic susceptibility fill. The fourth ditch is located to the west of it, the semi oval shape of the ditch can be recognized in the magnetic anomaly (see Appendix I for the interpretation diagram). Data range -2 to 4 nT; data has been interpolated. b: Two ditches at right angles, the clearest anomaly is caused by a north-south oriented ditch on the east side of the figure. The second ditch can be recognized in the bottom half of the figure, and has an east-west orientation. The reason why the first ditch causes a much clearer magnetic anomaly than the second can probably be attributed to the difference in depth (and volume) of the two ditches. The features were investigated by hand auger. The data was collected in Polre. Data range -7 to 7 nT; data has been interpolated. c: Two ditches with a slightly different orientation in Beugen, the first in the top of the figure, the second in the bottom. Interpretation has not yet been confirmed by excavation, and it is not clear if the proposed ditches belong to the settlement that was excavated in this area. Data range -2 to 3 nT; data has been interpolated.

98

7.1.3 Walls

Figure 48 Examples of the magnetic response of superficial and buried brick and limestone walls. a: The magnetic response of a brick wall that is visible on the surface, superimposed on the plan of the church of Valkenisse. There is a clear distinction between the area with and without brick. The magnetic response consists of many bipolar anomalies with random orientations, caused by the different directions of remanent magneti-zation in the individual bricks which are no longer in situ. The northern wall of the choir of the church shows a more consistent anomaly, apparently there is a dominant direction of magnetization in the bricks that make up the wall. Data range -6 to 10 nT. b: Three areas of buried brick walls or foundations in the drowned village of Polre, on the projected location of the church. All magnetic anomalies have a positive and a randomly oriented negative component. The first anomaly is located in the southeastern corner, the presence of solid brickwork was confirmed by means of hand auger. The second and third anomaly can be recognized as strongly positive anomalies with a negative compo-nent in the top half of the figure. Data range -7 to 7 nT; data has been interpolated. c: Two linear negative magnetic anomalies on a Roman villa site in Meerssen. The first anomaly is parallel to the grid edge, and approximately 4 meters away from it. The second negative anomaly is at right angles to it, in the bottom half of the figure. The interpretation of these anomalies remains speculative, but this is the type of response that can be expected from a buried limestone wall in a higher magnetic susceptibility matrix. Data range -2 to 4 nT; data has been interpolated.

99

7.1.4 Wells

Figure 49 An example of the magnetic anomaly caused by a buried well in Limmen. The positive anomaly that was caused by the well can be seen in the center of the dataset. For comparison, the much stronger positive anomalies along the top edge are caused by buried ditches. It is likely that the magnetic susceptibility enhance-ment of the well deposits has taken place in the deeper, primary deposits which are rich in organic matter. See Appendix I for more details. 7.2 Off-site 7.2.1 Plough marks

Figure 50 An example of striping in a magnetometer dataset, which is caused by ploughing. Plough marks are parallel to the edge of the magnetometer survey. Higher magnetic susceptibility material from the topsoil is ploughed into the lower magnetic susceptibility subsoil, resulting in linear positive magnetic anomalies in the location of the furrows. In this case the ploughing is modern, but ancient plough marks could have a similar magnetic response. Data from Borgharen.

100

7.2.2 Ditches

Figure 51 Examples of the magnetic anomalies that are caused by buried ditches. a: Harnaschpolder, a strong positive magnetic anomaly with a negative halo on the right hand side of the figure is caused by a post Medieval ditch which is buried at shallow depth. The feature was investigated by hand auger. The fill of the ditch has an enhanced magnetic susceptibility, and holds some remanent magnetic inclusions like pieces of brick. A second ditch can be seen in the northwestern corner of the figure, it appears to be less cohe-rent than the first ditch. It has not been investigated by hand auger. The striping which is visible in the anoma-lies is caused by stepping errors. Data has been low pass filtered. Data range -1 to 3 nT. b: Three, possibly four parallel ditches show as positive anomalies in data from Ossenisse. The ditches were first recognized on an aerial photograph, no intrusive investigations have been undertaken. Data has been inter-polated, the range is -2 to 4 nT. c: An example of the magnetic response of two field boundary ditches, the first ditch can be followed from the northwest corner in south-southwest direction, the second ditch is at right angles to the first in the bottom half of the figure. The location of the ditches is defined by the presence of bipolar magnetic anomalies, which are likely to be caused by remanent objects. The fill of the ditches may contain metal, brick or slag material. The response is much different and easy to distinguish from the geological responses which dominate the rest of the figure. The interpretation of the ditches is based on the historical map of 1832, which displays them. Data has been interpolated, the range is -1 to 2 nT.

101

7.2.3 Watering pits

Figure 52 An example of the anomaly that was caused by a buried watering pit, it consists of a positive mag-netic anomaly with a slight negative trough at its north side. The feature was investigated by hand auger, the enhanced magnetic susceptibility fill contained pottery and charcoal. The interpretation of the feature has been based on the results of the excavations in the immediate vicinity. Excavations that could confirm the inter-pretation have not yet been conducted. Details can be found in Appendix I. 7.3 Funerary structures 7.3.1 Graves

Figure 53 An example of the magnetic response of three, possibly four buried graves in Borgharen. The linear negative anomaly in the southwestern corner of the figure indicates the location of the trial trench in which a number of Medieval graves were uncovered. The shape, size and direction of the positive component of the magnetic anomalies that can be seen in the magnetometer data corresponds well to the shape, size and direction of the excavated graves. Two anomalies are located directly to the northeast of the trial trench, a third anomaly due north of the eastern anomaly. It is likely that the grave fill has an enhanced magnetic susceptibility, which is causing the magnetic anomaly. A possible fourth anomaly can be seen due east of the western anomaly, it appears to consist of two anomalies rather than one. Excavations that could confirm the interpretation of the data have not yet taken place. Data has been interpolated. More details can be found in Appendix I.

102

7.3.2 Tumuli (ring gullies)

Figure 54 An example of the magnetic response of a possible ring gully in Ossenisse. The circular positive magnetic anomaly is located in the northeastern corner of the dataset, and measures approximately 13 meters in diameter. Two triangular shaped positive anomalies on the northwestern side of the anomaly may indicate the location of a gap in the ring gully. A circular crop mark, which is part of a larger group of similar crop marks (see Appendix I) was seen in this location on aerial photographs. The interpretation of the features as funerary structures remains speculative until intrusive investigations have taken place. Data has been low pass filtered and interpolated. 7.4 Industrial 7.4.1 Peat ties and extraction pits

Figure 55 In Smokkelhoek, there is a magnetic difference between the presence and the absence of peat in the subsoil. Peat has a much lower magnetic susceptibility than the clastic sediment surrounding it. The peat has been partly extracted from pits. These pits have become filled in with clastic material, resulting in a positive magnetic contrast with the peat that is still in place. In the bottom left corner of the figure, linear negative magnetic anomalies can be observed, these are caused by in situ peat, whereas around these anomalies the peat has been extracted. This interpretation has been confirmed by hand augering.

103

7.4.2 Furnaces

Figure 56 Examples of the magnetic response of furnaces, which are usually characterized by strong positive anomalies with a strong negative halo or trough at the north side of the anomaly. a: Three furnaces have caused the magnetic anomaly that can be seen in the center of the bottom half the figure. It is not possible to recognize the three individual furnaces in the magnetic response, because the anomalies overlap. A fourth furnace has caused an isolated anomaly, it is located directly north of the other three furnaces. Data from Heeten. Excavation after the magnetometer survey has shown that the metal working furnaces are of the pit furnace type. Data has been interpolated, data range -3 to 3 nT. b: During the magnetometer survey in Valkenisse, remains of furnaces and metal working debris were observed in the stream bed of the tidal river Scheldt. These remains have caused a y-shaped positive anomaly with a negative halo surrounding it. One of the in situ furnaces is located at the northwestern tip of the anomaly, it appears to have caused a circular anomaly. No other individual features can be recognized in the magnetic response because the anomalies overlap. Data range -5 to 10 nT. c: Three possible furnaces in Polre. These strong positive anomalies with a strong negative component were interpreted as furnaces after evidence for metal working was found during hand augering over one of the anomalies. Data has been interpolated, data range -7 to 7 nT.

104

7.5 Infrastructure 7.5.1 Roads

Figure 57 Examples of the magnetic anomalies caused by buried roads and road side ditches. a: Two parallel road side ditches cause positive anomalies in the Medieval village of Limmen. The fill of the ditches has an enhanced magnetic susceptibility. The interpretation has been confirmed by excavation. Data range -2 to 4 nT. b: The remains of a road in Smokkelhoek are causing a positive magnetic anomaly. The location of the anomaly corresponds with the location of a road on the 1832 map. Data range -1 to 2 nT. c: In the data that was collected over the Roman Road in Swalmen, a linear negative anomaly can be seen (north – south direction). Intrusive investigations have not taken place, but it is likely that this anomaly corresponds to one of the road side ditches along the road. Data range -2 to 4 nT.

105

8 Discussion This discussion will start with the critical observation of the methodology that was used during the research that is reported on here, including site selection and instrument choice. The second section considers the factors that facilitate and hamper the formation and preservation of enhanced soil magnetic susceptibility in The Netherlands, which is of critical importance for the formation of induced magnetic anomalies. The topic will be discussed in conjunction with the results of the investigations that have been carried out on estuarine, wind blown and fluvial soils. Finally, the discussion will focus on the possibilities of magnetically mapping certain archaeological features and objects, unrelated to the geological environment in which they are embedded. 8.1 Methodology Chapter 4 discussed how, during the research, the methodology regarding site selection had diverted from the original research plan. Rather than investigating three large areas which included one or more archaeological sites, as was originally proposed, a larger number of smaller areas was investi-gated. Hence the research has concentrated on groups of individual sites rather than on the archaeo-logical landscape. The shift in the site selection process was made after magnetometer surveys had taken place in the first two larger research areas, where results were disappointing. It was thought that a larger diversity of sites was needed in order to investigate the differences between magnetic con-trasts in different areas and to understand the lack of contrast that was observed in the first two areas, Broekpolder and Harnaschpolder. On hindsight, the choice of these two sites, both on estuarine depo-sits, was unfortunate. Very few archaeologically relevant magnetic anomalies could be recognized in the datasets, but this lack of a magnetic contrast in archaeological deposits was observed on estuarine soils in general. In areas with another geological background the magnetic response was different, and in the Meuse valley, for example, investigating a larger area around an archaeological site would have probably proved worthwhile. Whether of not any of the elements that make up the (man-made) land-scape can be detected magnetically in the areas where there is an on-site magnetic contrast remains a topic of further research. The larger number of sites that was thus surveyed and otherwise investigated, allowed for a better overall interpretation of magnetic responses related to the different geological backgrounds in The Netherlands. Moreover, the new approach to site selection secured a much better ground truthing of the data. Using a small dataset like this (31 sites, two of which outside of The Netherlands), however, it remains difficult to generalize, but the geological approach has appeared to be useful to discuss similarities in magnetic susceptibility and magnetic response between the sites. Apart from the geolo-gical background, there are two other factors that may correlate to the presence of a magnetic contrast and the potential for a successful magnetometer survey; the age of the archaeological deposits and the type of archaeological site, which will be discussed in a later section of this discussion. The single type of magnetometer that was used during this research is the fluxgate gradiometer. The choice for a gradiometer as opposed to a total field magnetometer is obvious when the generally small magnetic variations that are caused by archaeological features are considered. As was discussed in Chapter 4, although the fluxgate gradiometer remains the magnetometer which is most used in archaeological prospection, other types, like the caesium vapour magnetometer are increasingly deployed. For a number of surveys that were conducted within the framework of this study, it is expected that the use of a more sensitive magnetometer, for example a caesium vapour instrument, would have yielded better results.

106

This is only the case, however, on sites where a magnetic contrast was present in the first place, like in Limmen and Beugen. Sites that lack this contrast, or sites on which the features have both a posi-tive and a negative contrast to the subsoil (see for example Fig. 32) would obviously not necessarily benefit from the use of a more sensitive magnetometer. Investigations into the iron mineralogy of estuarine deposits by means of Curie balance and IRM measurements, although not included in the original research plan, proved to be very important for understanding both geological and archaeological magnetic contrasts. If these magnetic measurements would have been part of the original research plan, a more systematic sampling strategy could have been applied and sites on wind blown and fluvial geologies could also have been investigated. 8.2 Contrasts in magnetic susceptibility This thesis has mainly been concerned with magnetic susceptibility contrasts within the three geo-genetic environments that were defined at the start of this study: estuarine, wind blown and fluvial. Different contrasts, for example between topsoil and subsoil, or between archaeological deposits and the subsoil, could be observed in these environments and in order to investigate the relation between the background geology and the possibilities for the magnetic mapping of archaeological features, the variables that influence the formation and preservation of magnetic contrasts will now be discussed. Organic matter The importance of the presence of organic matter for the enhancement of the soil magnetic suscepti-bility was discussed in Chapter 3. In the process of the transformation of non-ferrimagnetic compounds to ferrimagnetic iron oxides by heating, organic matter is needed to create reducing circumstances. For bacterial magnetic enhancement, the presence of organic material is important for two reasons; it acts as a nutrient for the bacteria, and it can facilitate reduction as a source of electrons. In the Netherlands, the magnetic susceptibility of topsoil material is generally enhanced. There are no sources for primary magnetite from a parent material. If the industrial fall out of ferrimagnetic particles is discounted, and there is no reason to believe that all topsoil enhancement can be attributed to this fall out (see Dearing et al. 1996), especially not below the surface level, than the observed enhancement is likely to be caused or facilitated by bacteria. If enhancement is so strongly related to the presence of organic matter, it can be expected that the fill of archaeological features, which usually have a higher organic matter content than the matrix that they are embedded in, is enhanced as well. Either because the 'archaeological' topsoil was already enhanced, and has collected in the fill, or because of the post depositional bacterial enhancement of the archaeological deposit. In a number of the case studies this appeared to be the case, in Meerssen for example, topsoil and feature fill magnetic susceptibilities are very similar. On other sites, samples from archaeological feature fills had higher susceptibilities than the topsoil, for example in Borgharen. In many of the case studies it was observed, however, that although the topsoil magnetic susceptibility on the archaeological site was enhanced when compared to the subsoil, in the archaeological features this enhancement could not be seen, or was less prominent. The formation of (ferrimagnetic) iron sulphides (for example in Smokkelhoek, Harnaschpolder and Spalding) appears to be linked to deposits which are rich in organic matter, suggesting a bacterial origin. In this study geological features in which these iron sulphides had been formed sometimes caused magnetic anomalies in the magnetometer data. In theory, the preferential formation of iron sulphides could also occur in archaeological deposits, provided that they have a higher organic matter content and that they are subject to sea or brackish waterlogged conditions. An example of preferential formation in organically rich archaeological deposits is the greigite which was en-countered in the postholes of the ‘seahenge’ monument (Linford 2006). During this study, the magnetic anomaly caused by the deep well deposits in Limmen may have been caused by the presence of iron sulphides in the deposits, but this remains speculative. Further research is needed to confirm the preferential formation of iron sulphides in archaeological deposits. Moreover, it may be difficult to distinguish the archaeological signal from the geological signal if iron sulphides are formed in both types of deposit.

107

Particle size Detailed investigations into the relation between magnetic susceptibility and grain size in archaeological deposits have been conducted by Weston (2002, 2004). He concludes from his experiments that finer soil fractions (clay and silt) are more easily enhanced, and that this enhance-ment is more difficult to undo by post depositional processes than in coarser material. Coarse soils without organic matter or a finer soil fraction can more easily be flushed of iron. Dearing et al. (1996, 2001) have suggested that ferrihydrite may be an important precursor of magnetite in English soils. Magnetite can be formed through bacterial reduction, during short periods with anaerobic conditions. This is best achieved in soils which have a large micropore volume on the one hand, but are free-draining on the other hand, identified as silty loams, clayey loams and silty clays, again the finer soil fractions. The lack of magnetic susceptibility enhancement in the fills of archaeological features on coarse sandy soils was also observed in this study, whereas the largest magnetic susceptibility contrasts occurred in silt, in clay and in sand and clay and sand and silt mixtures. An exception is the site of Limmen, were magnetic contrasts existed in dune sand, which has a smaller particle size than the cover sands in the eastern and southern part of the country. Soil pH If a soil is very acidic, the soil pH can influence the magnetic susceptibility of a deposit, whereas other pH values have very limited influence on the susceptibility (Weston 1999). In an acidic environ-ment, iron is more easily mobilized and can be flushed from or translocated within the deposit. In Broekpolder, the effect of an acidic top layer could be seen to have resulted in a low magnetic sus-ceptibility topsoil. Another example is the formation of podzols on for example the cover sands, where iron is flushed from the upper horizons and moved to a lower soil layer, the B-horizon. Although the iron is concentrated in this horizon, it is present as paramagnetic ferrihydrite. For this reason, magnetic susceptibility samples that were taken from a podzol in Den Dolder showed no enhancement for the B-horizon. The presence of a podzol soil is an indication of (past) acidic circumstances and the movement of iron within the soil profile. This process of podzolization will also affect any archaeological deposits, and will change their magnetic susceptibility. If calcareous material is present in a deposit, it can form a buffer for acidity, for example the shells in the estuarine soils. Focusing on the areas that have been studied, cover sands and Meuse valley deposits are non-calcareous, and more prone to magnetic susceptibility changes due to the loss of iron. Post depositional processes One of the post depositional processes that may affect the magnetic susceptibility of a deposit, podzo-lization, is discussed above. In a podzol iron is moved down the soil profile, but dissolution of iron can also occur in less well drained soils. Gleying is the dissolution and oxidation of iron due to the oscillating groundwater table. This does usually not affect archaeological levels, as these often occur above the groundwater table, but if it does, the magnetic susceptibility of the gleyed deposits will change. Mullins (1977) has suggested that a low magnetic susceptibility in gleyed soil layers is due to the dissolution of maghemite, but more generally, as a result of gleying, the iron oxide forms will have changed, and with it the soil magnetic susceptibility. Iron oxides may also change during water-logging, but the conditions will have to be long lasting and severe. A post depositional process that has received a lot of attention in this study is the so called sea waterlogging. During this process, soil iron oxides may dissolve to form iron sulphides. Again, the changes in the iron mineralogy affect the magnetic susceptibility of the soil. A complicating factor for magnetic prospection is the high mag-netic susceptibility of certain iron sulphides, most noticeably greigite, bodies of which can cause mag-netic anomalies that are much stronger than archaeologically caused anomalies. The iron sulphides that were observed in the field occurred in conjunction with organic matter, which would favour the interpretation of the iron sulphides as being bacterially mediated. 8.3 Estuarine and marine deposits The influence of the above factors on soil magnetic susceptibility, could be illustrated well on the sites of Broekpolder and Harnaschpolder, where more detailed magnetic measurements were conducted.

108

Organic matter content did not appear to bear a relationship to the soil magnetic susceptibility, archaeological feature fills, which were high in organic matter when compared to the matrix, did generally not have a magnetic contrast. This was found to be the result of several post depositional processes. One of these processes could be identified on both sites; gleying. During gleying, iron is mobilized and moved up in the profile, thus creating depleted and enriched areas. In such cases, fractional conversion values cannot be used to pinpoint 'archaeological enhancement', although they may be used to identify gleying in the profile, as was shown in Chapter 5. The present gley horizon may generally be visually identifiable, but evidence of past gleying may be much harder to define. In Broekpolder and Harnaschpolder the presence of oxidized magnetite (maghemite) and hematite in the subsoil provided additional proof for gleying. For archaeological prospection purposes, any magnetic contrast between the archaeological deposits and the matrix can generally be expected to have ceased in areas where gleying has occurred above the archaeological level. It is assumed that the influence of leaching could be seen in the data from Broekpolder, where the acidic upper part of the soil profile appears to be partly flushed of iron. Leaching, like gleying, translocates iron in the soil profile, and changes the iron distribution in the soil. It is interesting to note that neither gleying nor leaching could be observed in the soil magnetic susceptibility, as values were low, but homogeneous. Extremely acidic soil conditions can occur during the oxidation of a Potential Acid Sulphate Soil into an Actual Acid Sulphate Soil. It can be assumed that this acidity too affects the iron mineralogy of the soil. Non-oxidized iron sulphides in the subsoil occur in marine and estuarine deposits in The Nether-lands. On the one hand their presence causes large magnetic anomalies of a geological nature, often correlated to the presence of organic material in the subsoil. These anomalies can hamper the inter-pretation of magnetometer data which is collected for archaeological purposes. On the other hand, the preferential formation of iron sulphides may occur in archaeological deposits if their fill contains more organic material than the matrix that they are embedded in. If ferrimagnetic iron sulphides are present, archaeological feature fills may cause a magnetic anomaly. This hypothetical scenario needs to be tested on archaeological sites which are brackish or sea waterlogged. 8.4 Wind blown deposits The largest extend of superficial wind blown deposits in The Netherlands are the cover sands in the eastern and southern part of the country. There is generally a magnetic susceptibility contrast between the topsoil and the subsoil, which coincides with the boundary between subsoil deposits without much organic material content, and topsoil deposits which contain much more organic matter. The argument of the bacterial enhancement of the topsoil layer in the presence of organic matter could be valid for these deposits. Magnetic susceptibility contrasts between the fills of archaeological features and the matrix, however, could not be observed on archaeological sites on cover sands. Feature fills in this type of soil usually contain more organic matter than the surrounding matrix, a similar situation to the estuarine deposits. In this case, however, it is more difficult to determine if the magnetic susceptibility contrast has never existed or if it has existed but has been undone. The first reason is that detailed magnetic mineral investigations have not taken place, which makes it more difficult to reconstruct the magnetic history of a site. Moreover, many of the topsoil deposits in the cover sand area are man made, for which reason it is more difficult than in the case of the estuarine deposits to compare the present topsoil to the archaeological topsoil. One post depositional process can be visually identified, without magnetic measurements, and that is visible in much of the cover sand area is podzolisation. This process occurs readily because of the acidic nature of the soil environment and the coarse nature of the sand, which makes the soil prone to the leaching of iron. Leaching has possibly affected magnetic contrasts on archaeological sites on which podzolisation has taken place. This process may for example have caused the lack of or limited observed magnetic contrasts on the sites of Breda and Den Dolder. In Heeten, large magnetic susceptibility variations were observed within the lower soil layers of the podzol; the B-, BC- and C-horizon.

109

For the B- and BC-horizon these variations are likely to be specific for this site, because of the high temperature activities that have taken place, which may have caused local enhancement of the deeper soil layers. The observed C-horizon susceptibility variations, however, appear to be more general, and could be related to the presence of red sand. Where red sand occurs, distinguishing natural magnetic variations from the variations that are caused by archaeological features may be cumbersome. The influence of masking on the magnetic prospection of archaeological features is mentioned here, although it may occur in any of the geogenetic environments, because it is most relevant for plaggen soils. The presence of a masking layer will increase the distance between the archaeological feature and the magnetometer, which will result in the decrease of the magnetic anomaly strength. The influence of a masking layer on the resulting magnetic signal at the surface can easily be modeled. In contrast to the problems that were encountered on cover sands, the other two groups of wind blown deposits in The Netherlands, loess deposits and inland dunes, appeared to be far more suitable for magnetic prospection. On each of these geologies, only one archaeological site was investigated within the framework of this study. On these two sites, magnetic contrasts could be identified between the undisturbed matrix and the archaeological feature fills, and a number of archaeological features could be magnetically detected at the surface. More research is needed to assess the factors that have caused and preserved magnetic contrasts on these sites. The particle sizes of the coastal dune (fine sand) and loess (mainly silt) deposits when compared to the cover sands (coarse sands) may play an important role. 8.5 Fluvial deposits In fluvial deposits, particle size mixtures that favour the formation and preservation of magnetic contrasts (often containing clay and silt) predominate. In the Meuse valley, clear magnetic contrasts could be observed that resulted in the magnetic detection of archaeological features at the surface. In the central river area, the observations were more mixed, as certain features have a magnetic contrast to the undisturbed matrix, whereas others do not. More detailed investigations are needed in order to understand the formation, preservation and possibly deletion of magnetic contrasts in these deposits. 8.6 Magnetic anomalies unrelated to the geological environment Thermoremanent magnetic anomalies are formed by in situ heating of the soil, and include kilns, hearths and furnaces, of which only furnaces were mapped in this study. It is important to realize that the detectability of these features does not depend on the type of matrix that they are embedded in. The same applies to remanent magnetic objects, like bricks and metal working debris. In this study the difference in magnetometer response between walls or foundations (e.g. Valkenisse) on the one hand, and scatters of brick or pits filled with pieces of brick or pottery (e.g. Meerssen) could be clearly ob-served. Natural stone building material can have a magnetic remanence, this is the case in for example basalt and granite. The building material that was investigated in this study, limestone, did not have a magnetic remanence. Due to the very low magnetic susceptibility of the limestone, however, it is expected to cause a negative magnetic anomaly in most matrices.

110

111

9 Conclusion The aim of this study was to assess the possibilities of the application of magnetic methods for mapping and evaluating archaeological remains in The Netherlands. This conclusion sets out with a short summary of the principles that underlie magnetic archaeological prospection. It is followed by a description of the most important findings of this study and its implications for the use of magnetic methods in ARM in The Netherlands. 9.1 Principles of magnetic prospection The principle of magnetic prospection lies in the contrast between archaeological features and objects to the matrix that they are embedded in. This contrast can be remanent, being caused for instance by subjecting soil material to high temperatures, or induced, depending on differences in magnetic susceptibility. If there is a magnetic contrast, a magnetic field will be present around the feature or object. This magnetic field will interfere with the earth’s magnetic field and create anomalies in it. These anomalies can in theory be detected with a magnetometer, which is a passive instrument that can measure the local strength and direction of the earth’s magnetic field, but can not be adjusted to take measurements ‘at a certain depth’. In this study the archaeological feature is the smallest unit of investigation. Buried features with a magnetic contrast can obtain a magnetic field around them, which, depending on the volume, depth of burial and contrast of the features, can be measured on the surface. There are several processes that can cause a magnetic remanence in a material, but thermoremanence which occurs after a material has been heated to high temperatures, is the process that is most relevant to archaeological prospection. Features and objects with a remanent magnetization will always cause a magnetic anomaly in the earth’s magnetic field. In archaeological terms these features include brick walls and foundations, kilns, hearths, and feature fills that consist mainly of brick or tile. The only constrains for the magnetic detection of these features are their contrast, their size and their depth of burial. A material that has not been directly investigated in this study is natural stone with a magnetic remanence, for example basalt and tufa. Archaeological features made up from these materials are likely to cause a magnetic anomaly that may be detected in magnetic surveys, depending on their size and depth of burial. The certainty of the presence of a magnetic anomaly that is caused by the above mentioned features and materials alone could be an incentive to use magnetic methods more frequent-ly in archaeological prospection in The Netherlands. All materials can obtain an induced magnetization, which depends on the presence of an applied magnetic field like the earth’s magnetic field. All archaeological features can become magnetized in this way, and can cause an anomaly in the earth’s magnetic field. The strength of the induced magne-tization depends on the contrast in the magnetic susceptibility between the fill of negative archaeo-logical features like pits and ditches or of the material of which the feature consists like natural stone and the undisturbed matrix that they are embedded in. Whether of not this anomaly can be detected by means of magnetic methods is depending on its strength at the surface, which is related to the contrast in magnetic susceptibility and the depth, shape and volume of the feature. In this study it was shown that a wide range of archaeological features can be detected, for example pits, ditches, wells, walls and roads. Without detailed knowledge about the magnetic contrast, however, it is hard to predict which features can be detected.

112

The general detectability of archaeological features on a certain site can be partly predicted, however, because the formation, preservation and deletion of magnetic contrasts depend for a large part on the geological circumstances. This is the reason why for this thesis, the research area was divided into three parts. In order to assess the use of magnetic methods in The Netherlands for mapping archaeological features and sites, which was one of the objectives of this study, three geogenetic environments were considered; estuarine, wind blown and fluvial. It was found that for induced magnetic anomalies (caused by magnetic susceptibility contrasts) the magnetic contrast depend not only on material properties like grain size, organic matter content and pH, but also on post depositional processes, which may have altered the primary magnetic contrasts into the contrasts that we observe today. It must be stressed at this point that the magnetic detection of objects that have a remanent magnetism is independent to their geological environment. Furnaces and brick walls for example can be detected up to a certain depth, depending on their volume and the strength of magnetization, in any geological environment. The only post depositional process that could affect these objects would be mechanical weathering. The possibilities for the magnetic detection of induced magnetic anomalies will now be discussed per geogenetic environment. 9.2 Magnetic contrasts in different geogenetic environments Magnetometer surveys on estuarine deposits did not prove successful in mapping archaeological features. A number of younger features could be detected, but archaeological pits and ditches did not cause a magnetic anomaly. Neither could archaeological deposits clearly be distinguished from the undisturbed matrix based on their magnetic susceptibility values. It is thought that the magnetic simi-larity of the archaeological and the non-archaeological deposits is the result of post depositional processes like gleying, leaching, waterlogging and sea waterlogging, which is discussed in more detail below. Because of these conclusions, magnetometer surveys on estuarine soils are not recommended as an archaeological prospection tool. One exception is archaeological sites that are buried under saline or brackish groundwater, as the preferential formation of iron sulphides in organic deposits could make these sites very good targets. The Weichselien cover sands of the eastern and the southern part of the country is a further geological environment in which the use of magnetometer surveys in archaeological prospection is not re-commended. On the archaeological sites that were investigated in this study magnetic susceptibilities and contrasts were very low. The reason for the lack of differentiation in these sandy soils is likely to be found in the fact that magnetic susceptibility enhancement is difficult to achieve and easily lost in coarse grained sediments. Only one site was studied on loess and one on coastal dune deposits, both of which had very good magnetic susceptibility contrasts. The data that has been collected in this study is too limited to judge the suitability of these environments for the use of magnetic methods as an archaeological prospection tool, but the results are promising and these areas should certainly be selected for a follow up study. The fluvial environment gave mixed results. In the central river area sites with and without clear magnetic contrasts were observed, but further investigations are needed to assess what factors have enabled the formation, preservation and deletion of magnetic contrasts on these sites. For the Meuse valley, consistently good magnetic contrasts were detected, which makes this valley the most suitable starting point for the integration of magnetic methods into the current suite of prospection methods. 9.3 Masking and variability Two processes that may hamper the detection of any archaeologically relevant magnetic anomalies in any geological environment, masking and variability, are discussed here. Masking is the increase in distance between the surface (and the magnetometer) and the archaeological deposits by the deposition of sediments or other deposits over older archaeological levels.

113

This is common in fluvial, estuarine and marine areas where sedimentation is or has until recently been an on-going process. Peat and secondarily deposited wind blown sands can also have covered archaeological levels. Plaggen soils are man made soils, thick layers of organic soil superimposed on the Weichselien cover sands in the eastern and southern parts of The Netherlands. The presence of these soils over archaeo-logical sites on the one hand increases the distance between the archaeological levels and the surface from which the survey is conducted in a similar way to alluvial layers that can cover archaeological sites. On the other hand, the variation of magnetic susceptibilities and magnetic moments within the plaggen soil may be larger than the magnetic contrast between archaeological features and the cover sand underneath. In this case the archaeological features may have been magnetically detectable under a homogeneous cover, but they cannot be distinguished under a magnetically variable cover. The variation in magnetic susceptibility and magnetic moments within the undisturbed matrix in which the archaeological features are embedded can also hamper the detection of these features. An example of this is the occurrence of red sand in the eastern and central parts of the Netherlands, the magnetic contrast and the shape of the red sand features are similar to the contrast and shape that would be expected from archaeological features, which makes it difficult to distinguish between the two. Magnetic variations in the layers underneath the archaeological levels are well known from surveys over for example igneous geology in Ireland and the United Kingdom. The Netherlands lack super-ficial bedrock, but during this study it was found that large magnetic variations can also occur in unconsolidated, estuarine deposits. On the sites of Harnaschpolder and Smokkelhoek highly magnetic anomalies of a geological nature were mapped during a magnetometer survey. As the interpretation of magnetometer data is often only based on the shape and size of magnetic anomalies, these features were initially assumed to be man made. The magnetic investigation of these large magnetic variations however, showed that the features causing them were in fact natural, ferrimagnetic, deeply buried features. The presence of these could confuse the interpretation of the response of the archaeological features in the magnetometer data in a similar way as the red sand does, but the structures are gene-rally larger and more consistent and the magnetic contrast stronger. 9.4 Iron sulphide formation In this study the hypothesis is posed that one of the reasons for the observed lack of magnetic contrast on archaeological sites under marine or estuarine deposits in the Netherlands is caused by a soil chemical change during sea water logging. Magnetic contrasts in the soil depend mainly on the type and abundance of iron oxides. If a marine inundation takes place after an archaeological site has been abandoned, the iron oxides can dissolve if the sea water logging is persistent. The iron is liberated to form iron sulphides, accumulations of these can be preserved in anoxic conditions, causing the magnetic anomalies that were observed in the magnetometer data of the surveys carried out in the former estuaries. It is thought that any magnetic contrast that would have been present on these archaeological sites is ‘erased’ during this process, making magnetic methods unsuitable for pros-pection on archaeological sites with a post-abandonment estuarine phase, not only in The Netherlands but in general. For example, this study also incorporates an investigation into the magnetic properties of an archaeological site in the Fenlands, United Kingdom, where both the presence of highly magnetic iron sulphides and a lack of contrast in the archaeological deposits could be identified. In The Netherlands, most of the archaeological levels that were inundated with sea water during a marine transgression are presently above the groundwater table. The soil contains iron oxides most likely of a different type and abundance than before the marine inundation. The observation of younger features being magnetically clearer than older features on estuarine soils does fit into this hypothesis, the divide not being old-young but rather pre- and post-transgression. If valid, the hypothesis may in part explain the absence of a detectable magnetic contrast -both on- and off-site- as was observed on archaeological sites in Zeeland and Noord- and Zuid-Holland. The deletion of the archaeological magnetic contrast remains a hypothesis, however, and more research is needed to investigate the soil processes that take place during and especially after marine transgressions.

114

9.5 Returning to the objectives Now that the areas in which magnetic methods can and cannot be used as an archaeological pros-pection tool have been outlined, the other objectives of this study need to be re-entered into the discussion. The objective of assessing magnetic methods as a tool to map the archaeological land-scape has not been fully investigated because of an unfortunate choice of the first two, largest survey areas (see Chapter 8). A study into a quick soil sample based method in order to assess whether a magnetometer survey could prove successful in mapping archaeological features, the fourth objective of this study, has been conducted. The best sampling method has been used throughout this study, sampling archaeological and undisturbed deposits and measuring their magnetic susceptibility contrast is a good way of assessing the suitability of a site for magnetometer surveys, if such samples can be collected. The difference in magnetic susceptibility between the topsoil and the subsoil did not prove to be an indication of the presence of archaeologically meaningful magnetic contrasts. Almost all of the successful magnetometer surveys, i.e. surveys in which a number of archaeological features could be mapped magnetically, have had a high magnetic susceptibility topsoil, however, of more than 30 x 10-8 m3/kg. Topsoil magnetic susceptibility could be a rough indication of the suitability of a site for magnetic prospection if samples from the subsoil are not available. 9.6 The integration of magnetometry into ARM The knowledge that was gained during this study is aimed to be the basis for the integration of magnetometry into the archaeological prospection toolkit. In this conclusion the strengths and limi-tations of the methods have been described, and at this point a proposal for the phased introduction of magnetometer surveys in Dutch Archaeological Resource Management is proposed. Phase I - include magnetometry in the KNA as the method of choice for the prospection of metal working

sites and sites at which kilns or furnaces are expected to be present, - propose geophysical guide (leidraad geofysica) for the KNA, - magnetometer surveys to be conducted alongside other archaeological prospection methods in all

Meuse valley surveys in which archaeological remains are expected to be possibly present in the top meter of the soil.

In this first phase magnetometry is embedded in the framework of the KNA in which ARM work is usually conducted. Consistent magnetometer surveys in an environment that is expected to show good results can create a (local) base for the integration of magnetometry as a prospection tool. Phase II - magnetometer surveys to be conducted alongside other archaeological prospection methods in all

surveys on loess in which archaeological remains are expected to be possibly present in the top meter of the soil.

Phase III - magnetometer surveys to be conducted alongside other archaeological prospection methods in all

surveys on dune sands in which archaeological remains are expected to be possibly present in the top meter of the soil.

In phase II and III the application radius of the magnetometer is increased, again using areas in which good results are expected.

115

9.7 The application of magnetic methods The magnetometer is not a miraculous instrument that overshadows traditional prospection methods, but its strengths have to be recognized in order to strengthen the methodology of archaeological pros-pection as a whole. Its merits are not yet to be found in finding previously unknown sites but in assessing the value known archaeological sites. Used along side hand augering, magnetometry can obtain a much better resolution in the same time (approximately one hectare per day on a resolution of 0.25 x 1 meter). Another big advantage is that it maps archaeological features on a horizontal plane, ‘as a map’ as opposed to a vertical stratigraphy. For example the features of those archaeological sites that are found during a hand augering prospection can be made visible in a magnetometer survey. At present, traditional prospection methods like hand augering and test trenching are often used in cases where a magnetometer survey would produce better data as compared to the hand augering, and would be less intrusive as compared to the test trenching. It is in the investigation and validation of individual features and on known archaeological sites that magnetic methods can be readily employed On extensive sites magnetometer surveys can be used next to excavations in order to gain information from the unexcavated part of the site, or to plan future excavations. Archaeological or geological features like ditches and creeks that have sufficient magnetic contrast can be quickly mapped over great lengths. Modern or ancient disturbances in the archaeological record can be mapped magnetically in order to assess their size and location. Annual or bi-annual surveys on scheduled archaeological sites can be used to monitor any change or disturbance in the archaeological record. More work is needed to get a better insight into the application of magnetic methods in Dutch archaeology. The investigation of the iron mineralogy of archaeological features in fluvial environ-ments would be a logical continuation of this study. In conclusion, magnetic methods have proved to be a welcome addition to traditional prospection methods. It is not a magical method, it can be successfully used in some situations, but not in all. With this study it has become much clearer under which circumstances a magnetometer survey can be successfully applied, which will give a focus to the integration of this novel technique into the existing archaeological prospection toolkit.

116

117

Appendix I Fact sheets Introduction This appendix contains the factual data of the sites that have been investigated in this study. Every site description starts with a heading table which contains the geological and archaeological characteristics of the site, its location and details about any archaeological work that has been conducted either before or after the magnetometer surveys or magnetic susceptibility sampling. The code in the first line of the table is the internal project code. The central coordinates are Dutch grid (Rijksdriehoeksnet) in meters. ARCHIS numbers are ARCHIS ‘research numbers’, unless this number is not yet available, in that case the number for the proposed archaeological activity (‘onderzoeksmelding’) is used. These numbers have the prefix p. For archaeological sites without any archaeological activity, the observation number (‘waarneming’) is prefixed with o. The prefix m is used for scheduled archaeological monument numbers. The textual descriptions are short and aim only to present all the data that was collected at one site, this typically includes a set of magnetometer data and the results of any magnetic susceptibility measurements. The interpretation of the data in these fact sheets is on a basic level. Where relevant, the reader is pointed to the main text where the data is further interpreted and placed into a wider context. 1 Beugen Beugen Zuid II BE04 municipality Boxmeer central coordinates 192875, 409400 type of archaeological site settlement period MED lithology sand, silt depositional environment fluvial archaeological activity excavation (ARCHIS: 4450) executed by Amsterdams Archeologisch Centrum, Universiteit van Amsterdam (AAC-UvA) Introduction and aim During an excavation of a settlement and burial site in Beugen prior to development, soil samples were taken in order to investigate the magnetic contrast between the fill of the archaeological features and the undisturbed matrix. This excavation created the opportunity to sample well defined archaeo-logical features in addition to the archaeological layers that could be sampled by hand auger during the magnetometer survey that was carried out just to the east of this project (see 2 Beugen). Methodology All samples were taken from the feature fills and sections during the excavation of trench 16, directly west of the survey Beugen-Zuid I. In this trench only Medieval features were excavated. Apart from the topsoil material and the undisturbed matrix, two well fills and the core of a posthole feature were sampled, as was a modern feature that intruded into the archaeological levels. All features are from Medieval or later age.

118

Results The results of the magnetic susceptibility measurements are displayed in Table 1. Both topsoil samples have high values. This is also the case with the fill of the modern feature that was sampled, and it is likely that that the fill consists mainly of topsoil material. Magnetic susceptibility values for the undisturbed matrix are much lower, whereas the values for one of the well samples and the post-hole sample range in between the magnetic susceptibility of the topsoil and the undisturbed matrix. A decrease in magnetic susceptibility with depth can be seen in most soil profiles (see paragraph 3.6), and many archaeological features are magnetically detectable because they are (partly) filled in with higher magnetic susceptibility topsoil material, as is probably the case with the modern feature that was sampled. It is likely that to a lesser extend this is also the case for the other two archaeological features, they have a higher magnetic susceptibility than the undisturbed matrix at the same depth, which could make them magnetically detectable at the surface. The second well sample has a very low magnetic susceptibility and a slight negative contrast with the undisturbed matrix. An insufficient number of samples has been taken to draw any firmer conclusions on the possible detection of the archaeological features at this site. Table 1 The magnetic susceptibility of the soil samples from the excavation at Beugen, trench 16. sample interpretation magnetic susceptibility

x 10-8 m3/kg 1 feature fill MOD 21.34 2 well MED/PMED 17.47 3 undisturbed matrix 7.55 4 topsoil 29.34 5 topsoil 19.87 6 undisturbed matrix 6.24 7 well LMED 5.41 8 posthole core MED 11.70 2 Beugen Beugen-Zuid BM03 municipality Boxmeer central coordinates 192900, 408900 type of archaeological site settlement period IA, (ROM), MED lithology sand, silt depositional environment fluvial archaeological activity test trenches (ARCHIS: 3838) executed by Amsterdams Archeologisch Centrum, Universiteit van Amsterdam (AAC-UvA) Introduction and aim A rural area south of Beugen was investigated archaeologically by means of test trenches prior to development. Settlement traces from the Iron Age and the Middle Ages were found under a thin plaggensoil, including buildings, wells, ditches, pits and an oven (Langeveld et al. 2003). Some Roman Period features could also be mapped. A magnetometer survey following these excavations was carried out in order to assess if on this location magnetic methods could have (partly) replaced the intrusive investigations. Because the Medieval and the Iron Age features occur in separate clusters, there was an opportunity to investigate if there was a magnetic difference between the features from the two periods on a similar geological background. Methodology A fluxgate gradiometer survey was carried out on a resolution of 0.5 x 1 meter. Samples for magnetic susceptibility measurements were taken by hand auger in an east-west transect (Figs. 2 and 3). Two of the magnetic anomalies that were mapped were investigated by hand auger and sampled for magnetic susceptibility measurements.

119

Results The results of the magnetometer survey are displayed in Figure 1, and the interpretation diagrams for the data can be found in Figures 2 (western part) and 3 (eastern part). A number of magnetic anomalies can be distinguished, both linear and round or oval in shape. The linear structures have two orientations, both approximately SEE-NWW. The features of orientation 1 (Fig. 2 and 3) are perpen-dicular to the Medieval ditch that was excavated north of core A50. It is likely that the magnetic anomalies are caused by the presence of ditches in the subsoil. These ditches may be of Medieval origin. The features causing the anomalies of orientation 2 are also likely to be ditches. A core in one of these ditches (X2 in Fig. 3 and Appendix III) showed that the shallow archaeological feature was disturbed by ploughing, except for some brick - which may have been mixed in with the topsoil - no dateable material was present in the fill of the feature. Three oval positive anomalies can be recognized in the data with lengths of 4.5, 5.5 and 9 meter. The features that are causing them are likely to be large pits. Because of their large size it is unlikely that the anomalies have to be interpreted as wells. In trenches 5 and 40 (in the north and south of Fig. 3) two large Iron Age oval pits were excavated that were interpreted as cattle watering places. It is possible that the three magnetic anomalies are caused by similar features. Table 2 The magnetic susceptibility of the soil samples from Beugen-Zuid. The samples have been taken by hand auger. core depth interpretation magnetic susceptibility

x 10-8 m3 /kg A100 5-15 topsoil 25.86 A100 40-45 topsoil 24.58 A150 5-15 topsoil 24.01 A200 5-15 topsoil 39.31 A250 10-15 topsoil 38.15 A250 30-35 topsoil 30.88 A300 5-15 topsoil 34.90 A300 40-45 topsoil 30.24 A50 5-15 topsoil 30.52 X1 5-15 topsoil 28.54 X1 35-40 topsoil 34.11 A150 35-40 topsoil / arch. layer 26.62 A100 45-50 arch. layer 6.07 A150 50-55 arch. layer 11.35 A200 45-50 arch. layer 33.98 A300 50-55 arch. layer 14.12 A50 65-70 arch. layer 7.79 X1 100-110 arch. layer 15.57 X1 45-50 arch. layer 31.07 X1 50-55 arch. layer 35.70 X1 55-60 arch. layer 56.28 X1 60-70 arch. layer 36.17 X1 70-75 arch. layer 194.65 X1 75-80 arch. layer 29.95 X1 80-90 arch. layer 24.33 X1 90-100 arch. layer 13.52 A250 45-50 arch. layer? 19.88 A50 50-55 feature? 14.07 A100 65-70 undisturbed 4.91 A100 90-100 undisturbed 2.90 A150 55-65 undisturbed 4.41 A150 85-90 undisturbed 1.06 A200 100-105 undisturbed 1.00 A200 70-75 undisturbed 8.12 A250 90-95 undisturbed 1.28 A300 115-120 undisturbed 0.77 A300 70-75 undisturbed 2.78 A50 110-115 undisturbed 3.02 A50 75-80 undisturbed 4.36 X1 110-120 undisturbed 18.74 A50 95 Mn/Fe concretions 7.91

120

0 50 m

3 nT

-2 nT

192700/408745

193040/409025

A0

Figure 1 The results of the magnetometer survey in Beugen-Zuid. The results of the excavations have been superimposed on the magnetometer data. Core X1 (Fig. 3), in the largest of the three anomalies, showed that there is 50 cm of archaeological deposits present underneath the topsoil, with a large amount of pottery and charcoal. The pottery has been dated to the Iron Age. The two medium sized circular anomalies directly tot the west may be caused by ovens (excavated in the trench to the west) or by any type of pit, for example a water pit (two of which were excavated in the trench to the west of the features). Most of the small anomalies are bipolar, they may represent small pits, but they can also be caused by pieces of magnetic material like brick, metal or slag. After the excavation the trenches have been backfilled, the location of trenches 10, 13, 19 and 20, however, can still be recognized in the magnetometer data. A modern cable is running parallel to the road in the extreme west of the surveyed area. Details of the cores can be found in Appendix III. The magnetic susceptibility of the samples that were taken from the cores are displayed in Table 2, and the digested data in Table 3. Table 3 The mean, median, minimum and maximum values for four groups of magnetic susceptibility samples from Beugen-Zuid. N is the number of samples. mean magnetic susceptibility

x 10-8 m3/kg median x 10-8 m3/kg

minimum x 10-8 m3/kg

maximum x 10-8 m3/kg

N

topsoil 30.64 30.38 24.01 39.31 12 archaeological layer 16.74 14.10 6.07 33.98 8 feature X1 48.58 31.07 13.52 194.65 9 undisturbed 4.44 2.96 0.77 18.74 12

121

0 50 m

A0

? oven

A50

A100

A150

192890/408880

192700/408750

large pit, ? watering place

medium sized pit

? small pit

ditch - orientation 1

ditch - orientation 2

core with numberA0

Figure 2 Interpretation diagram of the surveyed area (western part). The location of the cores is indicated as well as the features from excavation that are referred to in the text. The magnetic susceptibility of the samples from the topsoil is very high compared to the undisturbed matrix. The magnetic susceptibility of the archaeological layer is very variable, and magnetic sus-ceptibility values are lower than the topsoil values. It may not be reasonable to consider the archaeo-logical layer as one homogeneous layer that is present over the whole surface of the site, but rather as a complex of archaeological deposits, the magnetic susceptibility values may reflect this in homo-geneousness. The deposits that are associated with the archaeological feature that was sampled, however, generally have higher magnetic susceptibility samples than both the topsoil and the subsoil samples. This con-firms that this archaeological feature could be magnetically mapped because of an enhancement in magnetic susceptibility of the fill of the feature. A difference in magnetic response between Medieval and Iron Age features could not be observed. A number, but not all, archaeological features could be mapped in Beugen. It is likely that some of these, because of their orientation, are Medieval, and some, for example X1, are Iron Age features. In this case a magnetometer survey would have given a quick indication of the location of certain fea-tures on the archaeological site, but because of the limited information that can be obtained from this data, it could not have replaced the trial trenching. A combination of hand augering and a magneto-meter survey, however, could in this case have provided information about the location and the dating of this previously unknown archaeological site.

122

0 50 m

A150

A200

A250

A300

X1

X2

192850/408770

193040/409040

? watering place

? watering place

water pits

? oven

Figure 3 Interpretation diagram of the surveyed area (eastern part). Key as in Figure 2. The location of the cores is indicated as well as the features from excavation that are referred to in the text.

123

3 Broekpolder Broekpolder BP02 municipality Heemskerk central coordinates 107600, 500800 type of archaeological site settlement, agricultural fields period BA, IA, ROM lithology clay, sand depositional environment estuarine archaeological activity hand augering (ARCHIS: 1036), test trenches (ARCHIS: p1072), excavation (ARCHIS:

p1855) executed by RAAP Archeologisch Adviesbureau; Rijksdienst voor het Oudheidkundig Bodemonder-

zoek (ROB); Amsterdams Archeologisch Centrum, Universiteit van Amsterdam (AAC-UvA)

Introduction and aim The site of Broekpolder is a scheduled archaeological monument, part of which has been integrated into a new housing development. Excavations around the current monument showed many archaeo-logical features connected to habitation, agriculture and religion ranging from the Bronze Age to the Middle Ages. A magnetometer survey was aimed to investigate which type of features from which period(s) could be magnetically detected. Magnetic susceptibility sampling, a heating experiment and IRM measurements were conducted to aid the interpretation of the magnetometer data. Methodology Magnetometer surveys were carried out on a resolution of 0.5 x 1 meter over a substantial part of the current monument. Two transects of cores (A and B) were conducted in order to obtain soil samples for magnetic susceptibility measurements (see appendix III for the hand auger data). The topsoil, presumed archaeological level, possible archaeological feature fills and the undisturbed matrix were sampled. An archaeological layer could often not be recognized, and in this case the material directly underneath the topsoil was sampled as if it were the archaeological level. After the low frequency magnetic susceptibility measurements, a selection of samples was subjected to a heat treatment in order to obtain the highest possible magnetic susceptibility for the specific soil material. Archaeological features were observed in the section of a freshly cut ditch, section A (Figs. 4 and 5). Samples were taken from the fill of these features, the topsoil and the undisturbed matrix for magnetic susceptibility measurements. The iron mineralogy of the soil samples from the section and of a modern ditch was investigated using an IRM component analysis. Results In the results of the magnetometer survey (Fig. 4) there are no magnetic anomalies that could be related to the archaeological features that had been excavated earlier in test trenches (overlay in Fig. 4). A zone of scattered magnetic noise can be seen behind the two entrances to the two fields; on the north side of the southern field and in the southwest corner of the northern field. Furthermore, large bi-polar magnetic anomalies (5 in the northern, and 1 in the southern field) are caused by the metal dipwells that were placed on the archaeological monument to monitor its conservation. Table 4 The magnetic susceptibility of the topsoil, the layer underneath the topsoil (assumed archaeological level), possible archaeological features and the undisturbed soil of soil samples taken by hand auger in Broekpolder. N is number of samples.



range x 10-8 m3/kg

topsoil 17.55 2.38 14.65 - 21.68 ‘archaeological level’, under topsoil 5.6 0.76 4.82 - 6.66 ?archaeological feature 10.47 4.67 7.02 – 21.90 undisturbed 4.98 1.72 2.98 - 9.80

124

In the magnetometer survey there is no indication of the magnetic detectability of archaeological features, but the transect magnetic susceptibility values (Table 4) suggest that the topsoil and subsoil material are magnetically different, and that the features that were recognized during hand augering have a fill with an enhanced magnetic susceptibility. It is possible, however, that these features are modern and not archaeologically relevant.

0 50 m

3 nT

-2 nT

107410/500660

107670/501020

A0 A18 A60 A80 A100 A120 A140 A160 A180 A200

B20 B40 B60 B80 B100

section A

SEA

Figure 4 The results of the magnetometer survey in Broekpolder. The location of the cores is indicated with a black dot and the coring number. Section A is located in the southwestern corner of the plan. The excavation trenches are superimposed on the magnetometer data.

125

Table 5 The results of the heating experiment, fractional conversion values for selected samples taken by hand auger. See Appendix II for the raw data of the experiment. level material fractional conversion % N Broekpolder topsoil sandy clay 3.27 2 modern ditch fill humic sandy clay 0.89 1 ‘archaeological level’, under topsoil sandy clay 0.48 2 ? archaeological feature sandy clay 0.83 2 undisturbed clayey sand 1.32 2 The heating experiment that was conducted on the augering samples resulted in very low fractional conversion values (Table 5), which indicates that the soil material has a good potential to become magnetically enhanced. There is sufficient iron present in the original soil, but it is present in a non-ferrimagnetic form. The only consistent linear magnetic anomaly that can be seen in the magnetometer data is located in the southern field (SEA in Fig. 4). It could not be related to any of the known archaeological features. Hand augering and subsequent low frequency magnetic susceptibility measurements showed that a filled in modern ditch was present on this location, and that its high magnetic susceptibility fill was embedded in a low susceptibility matrix, this being the likely cause of the anomaly. In contrast to this enhanced magnetic susceptibility was the lack of an elevated magnetic susceptibility in the fill of the archaeological ditches that were observed in section A (Fig. 5 and Table 6). The topsoil magnetic susceptibility in the section is higher than the subsoil susceptibilities, but the values of the fill of the archaeological features is not consistently higher or lower than the magnetic susceptibility of the matrix that they are embedded in. Moreover, the features are not visible in the results of an additional magnetometer survey (not displayed). An IRM component analysis was carried out in order to shed light on the mineralogical difference between the fill of the high magnetic susceptibility modern ditch SEA, that caused a detectable magnetic anomaly, and the fill of the archaeological ditches that did not appear to be magnetically different from their surrounding matrix from section A. The results of the IRM measurements are displayed in Table 7. Based on the IRM measurements, the iron mineralogy of the undisturbed (subsoil) samples on this site appears to be very similar to that of the fill of the archaeological features, both contain oxidized magnetite (possibly maghemite) and hematite in comparable ratios. The topsoil and the fill of feature SEA, however, are different, and both contain less oxidized magnetite next to hematite. In SEA, the contribution of magnetite to the signal appears to be greater than in the topsoil samples. These measurements confirm that on the one hand the modern feature is filled with topsoil-like material, which can explain the magnetic anomaly that is apparent in the results of the magnetometer survey. Table 6 The magnetic susceptibility of the samples from the ditch section in Broekpolder. See Figure 5 for the location of the samples. sample interpretation magnetic susceptibility

x 10-8 m3/ kg 1 topsoil 18.68 2 undisturbed 5.84 3 undisturbed 6.22 4 topsoil 21.13 5 feature 4.83 6 feature 6.78 7 topsoil 19.61 8 feature 7.73 9 feature 3.78 10 topsoil 17.76 11 arch. layer 6.45 12 undisturbed 8.49 13 topsoil 9.92 14 arch. layer 5.92 15 undisturbed 4.75 16 topsoil 8.18 17 arch. layer 5.40 18 undisturbed 5.75

126

On the other hand, the iron mineralogy of the archaeological features that were sampled in the ditch section is very similar to the mineralogy of the undisturbed matrix, and this lack of contrast may be the reason that no detectable magnetic anomalies have been caused by archaeological features. A more detailed discussion of the laboratory results can be found in Chapter 5. Table 7 Results of the component analysis of the IRM data of the ditch section samples from Broekpolder. The value for the SIRM depends on the amount of magnetic material that is present in the sample, B1/2 is independent of the quantity of material and a guide for the type of material (see Kruiver et al. 2001). DP is the dispersion factor, indicative for the width of the distribution. Low DP’s occur in crystalline material. Refer to Figure 4 for the location of sample SEA, and to Figure 5 for the remaining samples.

sam

ple magnetic

susceptibility x 10-8 m3/kg

mass x 10-3 kg

description

com

pone

nt B1/2

mT SIRM A/m

SIRM x10 -3 Am2/kg

SIRM/� A/m

DP interpre-tation

contribution to signal %

1 18.68 0.50 topsoil 1 42.2 0.78 15.6 83511.78 0.39 magnetite 93 2 631.0 0.055 1.1 5888.651 0.22 hematite 7 2 5.84 0.52 undisturbed 1 56.2 0.137 2.634 45102.74 0.46 oxidized

magnetite 91

2 707.9 0.014 0.2692 4609.589 0.30 hematite 9 3 6.22 0.50 undisturbed 1 56.2 0.132 2.64 42443.73 0.40 oxidized

magnetite 86

2 707.9 0.022 0.44 7073.955 0.37 hematite 14 4 21.13 0.47 topsoil 1 42.7 0.88 18.723 88608.61 0.35 magnetite 91 2 501.2 0.085 1.8085 8558.921 0.47 hematite 9 5 4.83 0.49 feature 1 53.7 0.124 2.5306 52393.37 0.42 oxidized

magnetite 94

2 501.2 0.0085 0.1735 3592.133 0.30 hematite 6 6 6.78 0.51 feature 1 52.5 0.106 2.0784 30654.87 0.38 oxidized

magnetite 91

2 501.2 0.0105 0.2059 3036.873 0.48 hematite 9 SEA 10.55 0.50 modern

feature 1 41.7 0.555 11.1 105213.27 0.38 magnetite 99

2 631.0 0.03 0.6 5687.20 0.2 hematite 1

Figure 5 Section A, section in a freshly cut ditch; a: two ditches are clearly visible in the section; b: schematic representation of the ditch section, location of photograph in grey. Numbers are sample numbers.

127

4 Deil Deil – Enspijk, A2 exit 15 DE04 municipality Geldermalsen central coordinates 142616, 432340 type of archaeological site ? settlement period BA lithology clay, silt, sand depositional environment fluvial archaeological activity excavation (ARCHIS: 7756 ) executed by ADC Archeoprojecten Introduction and aim Prior to infrastructural development, coring showed a possible Bronze Age settlement on this location, but no prehistoric archaeological features were found during the excavation of consecutive test trenches. A vegetation layer, possibly indicative of the Medieval surface layer, however, could be identified and samples for magnetic susceptibility measurements were taken in order to investigate if this layer was magnetically anomalous. Two possible Medieval pits that were buried under the vege-tation layer were also sampled. Methodology Samples were taken directly from the excavation trenches in four categories; topsoil, Medieval surface, Medieval pits and undisturbed matrix (C-horizon). It was thought that samples taken next to the A2 motorway could have a magnetic susceptibility enhancement that is caused by the fall out of air borne magnetic particles. Results The results of the magnetic susceptibility measurements are displayed in Table 8. The values in all four categories are fairly low. The magnetic susceptibility of the topsoil samples is slightly higher than that of the samples that were taken from the undisturbed matrix, but both the archaeological features and the vegetation layer have low susceptibilities. It is unlikely that the archaeological fea-tures that have been sampled would cause a magnetic anomaly that could be detected in a magneto-meter survey, neither does the vegetation layer seem to be magnetically anomalous. There is no indi-cation that the samples have been enhanced due to the fall out of the motorway. Table 8 The magnetic susceptibility of the soil samples from Deil – Enspijk.

ssample interpretation magnetic susceptibility x 10-8 m3/kg

1 topsoil 9.49 2 topsoil 7.48 3 topsoil 11.96 4 vegetation horizon (? MED surface) 5.14 5 vegetation horizon (? MED surface) 5.29 6 vegetation horizon (? MED surface) 5.04 7 natural 5.02 8 natural 4.06 9 natural 5.04 10 archaeological feature (?MED) 5.43 11 archaeological feature (?MED) 3.76

128

5 Den Dolder Den Dolder - Hezer Eng DO04 municipality Zeist central coordinates 145924, 461614 type of archaeological site grave field (BA), settlement (MED) period BA, MED lithology sand, boulder clay depositional environment aeolian over glacial archaeological activity excavation (ARCHIS: 7889) executed by Rijksdienst voor Oudheidkundig Bodemonderzoek (ROB) Introduction and aim Soil samples for magnetic measurements were taken during the excavation of a Medieval settlement and a group of Bronze Age tumuli, in order to assess the contrast in magnetic susceptibility between the features from both periods and the matrix that they have been cut into. At the time of sampling excavations concentrated on trench 1, the excavations were later expanded. Methodology Soil samples were collected, stored and measured as described in Chapter 3. Geologically, the site consists of wind blown sand that is deposited on top of boulder clay of a glacial origin. In the soil which has developed in the cover sand, the C – BC – B stratigraphy is in tact, the E-horizon has been damaged during ploughing and the A-horizon has disappeared. Samples 1 to 8 (Fig. 6) are background samples and are aimed to investigate the variation in magnetic susceptibility throughout the soil section.

Figure 6 Location of the samples for magnetic susceptibility measurements. Excavation plan of ROB, trench 1 level 1. Numbers are sample numbers.

129

Samples 9-15 were taken from the ditch fill of one of the tumuli and its interior, which consisted of boulder clay. A further sample of undisturbed boulder clay and a sample of the topsoil over the tumulus were also included. The fill of two medieval ditches (S24 (some burnt material was noticed) and S28) was sampled, as was S27, a pit feature of which the lower fill consisted of burned material. Finally, samples were taken of a number of posts and posthole fills form a Medieval hay stack feature (sample 25 to 30). The location of sample 31 is unknown. Results Background samples The magnetic susceptibility varies from 0.97 to 13.33 x 10-8 m3/kg (Table 9). The magnetic susceptibility of the topsoil samples is slightly higher than that of the subsoil samples that were taken from the section; except for sample 6, magnetic susceptibility decreases with depth. Tumulus samples Samples from the ring ditch have a magnetic susceptibility comparable to the magnetic susceptibility of the B- and C-horizon in the background samples, the contrast is not large enough for the ring ditch to be detected in a fluxgate gradiometer survey. The charcoal stain in the tumulus (in later excavations this feature turned out to be a well) does have a clear magnetic contrast to the matrix. The boulder clay on the interior of the tumulus has a contrast to the sand of the matrix around it. It is not clear if the tumulus was constructed of boulder clay. If this were the case, the contrast is not high (less than 10 x 10-8 m3/kg) but because of the volume of the body of the tumulus the feature could probably have been easily magnetically detected. Medieval feature samples Most samples vary in magnetic susceptibility between 0 and 2 x 10-8 m3/kg, these values are very similar to the background values that were obtained, and it can be assumed that the magnetic contrast is not large enough to be detected with a conventional fluxgate gradiometer. Exceptions are samples 18 and 19, both from S27, in which the presence of burnt material was noticed. These samples have a magnetic contrast and the feature it was obtained from would possibly have been detected in a fluxgate gradiometer survey. Table 9 Magnetic susceptibility of the soil samples from Den Dolder – Hezer Eng. sample interpretation magnetic susceptibility

x 10-8 m3/kg 1 topsoil 7.37 2 ploughed E 5.05 3 B 2.11 4 BC 2.33 5 topsoil 6.36 6 ploughed E 9.90 7 B 2.68 8 C 1.34 9 ring ditch, BA 2.15 10 ring ditch, BA 1.91 11 ring ditch, BA 1.92 12 tumulus interior (boulder clay) 2.18 13 boulder clay natural 2.80 14 tumulus interior (boulder clay) 2.92 15 topsoil over tumulus 9.38 16 trench 1, S24, burnt material, MED 0.97 17 trench 1, S24, MED 1.98 18 trench 1, S27, burnt material, MED 13.33 19 trench 1, S27, burnt material, MED 6.61 20 trench 1, S27, MED 2.13 21 trench 1, natural B, boulder clay 3.52 22 trench 1, natural BC 2.51 23 trench 1, S28, ditch, MED 1.23 24 trench 1, S28, ditch, MED 1.06 25 trench 1, S46, posthole 0.46 26 trench 1, S46, posthole 1.68 27 trench 1, natural 0.53 28 trench 1, S43, posthole, MED 1.48 29 trench 1, S72, posthole, MED 1.59 30 trench 1, S72, posthole, MED 0.85 31 charcoal stain in tumulus, BA 4.97

130

6 Breda Breda Wolfslaar GI03 municipality Breda central coordinates 114130, 397150 type of archaeological site settlement (IA), off-site (MED), battle-field (PMED) Period IA, MED, PMED Lithology sand depositional environment aeolian archaeological activity excavation (ARCHIS: 4355) executed by Brandenburgh Introduction and aim An excavation was conducted prior to the widening of a brook as a nature development on the edge of a Pleistocene sand ridge. An aerial photograph of the area showed linear crop mark features that could possibly be associated with the siege of Breda in 1624/25. A magnetometer survey was carried out in order to map any features that were associated with these crop marks. Methodology A grid of 0.5 x 1 meter was used for the magnetometer survey. Four soil samples for magnetic susceptibility measurements were taken by hand auger on the location of a magnetic anomaly. Further soil samples for magnetic susceptibility measurements were taken from excavation trench 1 (marked WP1 in Fig. 7), which was located 135 meter north of the geophysical survey area. In this trench Iron Age and Medieval archaeological features were sampled that were probably unrelated to the crop marks on the aerial photograph. The results of the magnetometer survey and of the magnetic susceptibility measurements will hence be discussed separately. Results Magnetometer survey Figure 7 displays the results of the magnetometer survey and the location of the excavation trenches, superimposed on the aerial photograph showing the crop marks that could possibly be related to the 17th century siege of Breda. A clear zone of magnetic noise can be seen in the magnetometer data. The anomaly is magnetically inhomogeneous, and is likely to be caused by remanent magnetic objects in the subsoil, for example pieces of metal, brick, or magnetic pebbles. The nature of the anomaly could not be identified in a core that was taken over the anomaly. Neither did magnetic susceptibility values of the sediment (Table 10) show any enhancement on this location. A small test pit was dug in order to clarify the cause of the anomaly, but in the wrong location (WP4 in Fig. 7). In none of the trenches that crossed the crop marks any archaeological correlate to the features could be found. In Figure 8, the location of the crop marks and the magnetic anomaly have been indicated. The area of magnetic noise does appear to be connecting to the main set of crop marks (in solid white line), and seems to be parallel to an earlier crop mark (hatched white line). This latter crop mark is over cut by the main set of crop marks, and is also parallel to the current field boundary. The interpretation of the crop marks, the magnetometer data and their relation is problematic. The crop marks may be very recent tunnels in the crop, made by children (pers. comm. R. Berkvens) but their structure seems to be very straight and complex. Alternatively, the crop marks may relate to soil features that have different physical properties than their surrounding matrix, but are not visible as archeological features, for example because they are concentrated in the plaggensoil. Magnetic susceptibility measurements The results of the magnetic susceptibility measurements are displayed in Table 10, apart from samples 6, 1, 15 and 16, that have been discussed above, all samples were taken from an excavation trench. The topsoil values are much higher than the values of the subsoil (C-horizon) samples, and the latter appear to have a very low magnetic susceptibility. The fill of the archaeological features (Iron Age and Medieval) have very similar, very low magnetic susceptibility values.

131

The archaeological features that have been investigated would not have caused a detectable magnetic anomaly, due to the lack of, or very limited contrast between the fill of the archaeological features and the matrix they are embedded in, in combination with a higher magnetic susceptibility, variable top-soil.

Figure 7 The results of the magnetometer survey in Breda superimposed on the aerial photograph with crop mark features. Data has been interpolated. The outline of the excavation trenches is indicated in white with WP for trench plus number.

132

Table 10 The magnetic susceptibility of the soil samples from Breda Wolfslaar. Samples 6,1,15 and 16 were taken by hand auger, the remaining samples from excavation trench 1. sample interpretation magnetic susceptibility

x 10-8 m3/kg 6 magnetic anomaly - topsoil (30-40 cm) 0.92 1 magnetic anomaly - under topsoil (50-60 cm) 0.54 15 magnetic anomaly - under topsoil (70-80 cm) 0.38 16 magnetic anomaly – C-horizon (80-90 cm) 0.69 3 topsoil 16.17 4 topsoil 8.72 5 topsoil 13.04 11 topsoil 13.45 7 archaeological feature 0.82 10 archaeological feature 1.01 8 archaeological feature with charcoal 1.34 14 archaeological feature with charcoal 0.79 13 C-horizon 0.12 12 C with Fe staining 1.02 2 C-horizon 0.15 9 C-horizon 0.22

Figure 8 The results of the magnetometer survey superimposed on the aerial photograph. Data has been interpolated. The crop marks have been traced with white solid and hatched line. The magnetic anomaly is indicated with a solid grey box.

133

7 Borgharen Borgharen Pasestraat 6 GM03 municipality Maastricht central coordinates 176300, 321700 type of archaeological site villa (ROM), grave field (EMED) period ROM, EMED lithology silt, sand, clay depositional environment fluvial archaeological activity excavation (ARCHIS: 2158), hand augering (ARCHIS: 10494), test trenches

(ARCHIS: 11440) executed by Gemeente Maastricht, RAAP Archeologisch Adviesbureau, Becker en Van de

Graaf Introduction and aim The Roman villa on Pasestraat 6 in Borgharen was partly excavated in two test trenches in 1999. The villa was in a poor state of preservation. Apart from the features that were associated with the Roman villa, a number a Merovingian graves were excavated. A magnetometer survey was carried out in order to assess if features associated with the Roman villa could be detected with magnetic methods, and if any other archaeological features were present. This work was carried out within in the framework of a much wider project, prior to large scale development in the area. It was not possible to carry out magnetometer surveys on other locations, but samples for magnetic susceptibility analysis could be taken from some of the archaeological features that were excavated. Excavations in the area where the magnetic survey was conducted are planned at a later stage. Methodology A magnetometer survey was conducted on a resolution of 0.5 x 1 meter. Samples for the background magnetic susceptibility values were taken from section Borgharen 210 (inspection hole 82). An un-dated oven was sampled in a test trench to the northeast of the surveyed area. In trench 11, samples were taken from two pits from the Iron Age. Results The results of the magnetometer survey are displayed in Figure 9. The area under investigation appears to be magnetically very noisy, this noise could be caused by pieces of brick and roof tile that could be seen on the surface of the ploughed parts of the field during the survey, or because of the presence of magnetic pebbles in the river sediments. In Figure 10 the location of the test trenches of 1999 is indicated with a grey fill, the features within with a black line. The location of the main test trench is also clearly visible in the magnetometer data in Figure 9. The direction of ploughing can be seen in the magnetic data, and for reasons of clarity, only part of these responses are indicated in Figure 10. On the east side of the survey a highly magnetic anomaly is probably caused by a modern pit (because of its location next to the road, pers. comm. former farmer of the land). In the northern part of the survey there are many anomalies that could be caused by pit fills, there are also some anomalies that are caused by remanent magnetism, most likely pieces of metal. In line with the excavated Merovingian graves, four magnetic anomalies that are likely to represent similar graves were detected. To the south east of the grave area, a piece of a hypocaust floor that was constructed from brick elements was excavated. A number of large, highly magnetic anomalies are visible near the location where the floor was excavated, these could indicate the presence of more building material, but they may also be caused by further grave fills. On the south western edge of the surveyed area a substantial linear magnetic anomaly was mapped that could possibly be related to part of a (brick) building, or to a feature that has been subjected to intense heating. Parallel to this feature, to the south of it, a linear area of increased magnetic noise can be seen that could be an indication of the presence of further building remains in this location.

134

The results of the magnetic susceptibility measurements are displayed in Table 11. The subsoil samples of section 210 all have reasonably low magnetic susceptibilities between 5 and 16 x 10-8 m3/kg. The topsoil values for section 210, and for the other topsoil samples are enhanced and all exceed 30 x 10-8 m3/kg. The highest magnetic susceptibility values were measured in the archaeo-logical samples, both the oven and the Iron Age pits have very high magnetic susceptibilities, which indicates a good magnetic contrast between the archaeological features and the undisturbed matrix. It is likely that the features that have been sampled would have caused a detectable magnetic anomaly in a magnetometer survey.

0 50 m

3 nT

-3 nT

Pases

traat

176400/321800

176200/321550 Figure 9 The results of the magnetic susceptibility survey in Borgharen, Pasestraat 6. The data has been inter-polated. The location of the main test trench is clearly visible, as is the direction of the ploughing.

135

0 50 m

Pases

traat

176400/321800

176200/321550

graves

?building

ploughing

?pits

metal

metal

metal

?pits

?pits

?pits

modern pit

hypocaust floor

Figure 10.8.2

Figure 10 The location of the 1999 test trenches (grey fill) and the features within (black line). The location of the observed magnetic anomalies is shown as black polygons with a grey fill. For reasons of clarity, not all plough marks are displayed. Table 11 The magnetic susceptibility of the samples from Borgharen. Note that all samples are taken from excavations to the east and the south of the surveyed area. sample description magnetic susceptibility

x 10-8 m3 / kg 1 oven 1 191.48 2 oven 2 172.44 3 oven 3 108.50 4 topsoil 1 44.79 5 topsoil 2 42.00 6 topsoil 3 39.35 7 section 210 – topsoil 33.85 8 section 210 – layer 2 11.81 9 section 210 – layer 3a 15.20 10 section 210 – layer 3b 11.19 11 section 210 – layer 4 8.33 12 section 210 – layer 7 5.65 13 section 210 – layer 8 10.50 14 trench 11 feature 257 213.14 15 trench 11 feature 253 182.71 16 trench 11 feature 253 189.66

136

8 Harnaschpolder Harnaschpolder HP02 municipality Midden-Delfland central coordinates 81600, 447900 type of archaeological site off-site period ROM lithology clay, silt, sand, peat depositional environment estuarine archaeological activity coring (ARCHIS: 417), test trenches (ARCHIS: - ) executed by RAAP Archeologisch Adviesbureau, Amsterdams Archeologisch Centrum, Uni-

versiteit van Amsterdam, (AAC-UvA) Introduction and aim The lay out of the Harnaschpolder Iron Age / Roman Period parceling system is well known after it has been investigated by means of test trenches. A large scale magnetometer survey was carried out in order to test the magnetic detectability of these off-site structures. The magnetic anomalies that were mapped during the survey were investigated with magnetic measurements. Additional sampling was carried out on two further archaeological excavations to the north and to the south of the area under investigation (see 9 Harnaschpolder north, east and south). Methodology A magnetometer survey was conducted on a spatial resolution of 0.5 x 1 meter. Four coring transects were made through the area, A, B, C, and D (Fig. 12). Soil samples for magnetic susceptibility were taken from the auger. In transect A and B four samples were taken from each core, topsoil, archaeo-logical level, undisturbed material and undisturbed at one meter depth. Generally, an archaeological layer could not be distinguished and the layer directly below the topsoil was sampled as being the archaeological level, for this is the layer in which archaeological remains from the Iron Age / Roman Period would be expected. The hand augering data can be found in Appendix III. A heating experiment has been carried out on a selection of the samples (Appendix II). Thermomagnetic measurements (Appendix II) have been conducted on samples that were collected separately from core PYR (Fig. 12) and freeze dried immediately. Results The results of the magnetometer survey (Fig. 11) show a number of very clear magnetic anomalies. Throughout the area under investigation anomalies that are caused by (ferromagnetic) metal occur and in the western part of the survey area some very strong alternating positive and negative magnetic responses are caused by the presence of subsoil pipelines. On the east side, strong magnetic responses relate to the remains of a demolished greenhouse on this location. In the central part of the survey area two types of response can be seen, well defined, negative anomalies with a creek like appearance and a group of mainly positive anomalies that has less well defined edges. The interpretation of these two groups of anomalies is based upon the results of the hand augering and the laboratory measurements. The creek like structures are likely to date from after the peat formation, the anomalies are caused by the very low magnetic susceptibility of a layer of silt that was sampled in B20 and that has cut into the top of the layer of Hollandpeat (e.g. in B40). Table 12 The magnetic susceptibility of samples from transect A. N is number of samples. Samples that deviate from the mean by more than 5 standard deviations have been excepted, these are topsoil A140 (181.86 x 10-8 m3/kg) and archaeological level A0 (40.16 x 10-8 m3/kg). These samples are assumed to have very magnetic inclusions like metal, brick or hard coal.



range x 10-8 m3/kg

N

topsoil 12.88 2.69 8.65 - 19.30 14 ‘archaeological level’, under topsoil

9.81 1.83 6.79 - 12.60 12

undisturbed 7.05 1.67 3.28 – 9.03 11 undisturbed 100 cm depth 5.19 2.41 0.48 – 9.23 15

137

The positive magnetic anomalies are caused by the presence of ferrimagnetic iron sulphides in the layer underneath the Hollandpeat. These iron sulphides, most likely a mixture of greigite and non ferrimagnetic pyrite, were identified from core PYR by means of thermomagnetic measurements (see Chapter 5 for a detailed description). Two post Medieval archaeological features were magnetically defined in the central part of the sur-vey. Both create a strong positive magnetic anomaly. In A0, a layer with a high magnetic susceptibi-lity was identified just underneath the topsoil, containing fragments of hard coal and brick.

Figure 11 The results of the magnetometer survey in Harnaschpolder (top) and the interpretation of the magne-tic anomalies (bottom). Data has been high pass filtered in order to remove the influence of the greenhouses to the south of the survey area.

138

Table 13 The results of the heating experiment on selected samples from transect A and B. sample depth interpretation magnetic susceptibility

before heating experiment x 10-8 m3/kg

magnetic susceptibility after heating experiment x 10-8 m3/kg

fractional conversion

A40 30-35 under topsoil 9.46 194.63 4.86 A60 5-15 topsoil 14.18 288.21 4.92 A60 30-40 under topsoil 7.95 221.57 3.59 A80 30-40 A-horizon 10.16 255.28 3.98 A80 30-40 A-horizon 8.51 316.15 2.69 A120 25-45 feature 13.64 224.51 6.08 A120 25-45 feature 11.77 209.46 5.62 A140 25-35 under topsoil 9.21 376.66 2.45 A180 5-15 topsoil 13.50 376.77 3.58 A180 35-40 A-horizon 11.56 453.96 2.55 A180 50-60 undisturbed 6.65 114.23 5.82 A180 50-60 undisturbed 7.03 132.70 5.29 A300 45-50 undisturbed 4.48 824.48 0.54 B20 99-102 undisturbed 0.00 2578.82 0.00 B40 40-45 undisturbed 7.78 2539.46 0.31 B40 40-45 undisturbed 7.63 2372.97 0.32 The earlier archaeological features that were found in the test trenches, the Roman Period parceling ditches for example, could not be magnetically mapped. Fractional conversion measurements have shown that although the magnetic susceptibility of the soil layers is low (Table 12), the soil layers have a potential for obtaining a high magnetic susceptibility (Table 13). An investigation of the magnetic contrast between the fill of archaeological features and the matrix they are embedded in was carried out with soil samples from Roman Period excavations to the north and to the south of the survey area (see 9 Harnaschpolder north, east and south).

Figure 12 The results of the magnetometer survey superimposed with the locations of the cores.

139

9 Harnaschpolder north, south and east Harnaschpolder north HP03N municipality Midden-Delfland central coordinates 81620, 448320 type of archaeological site settlement period ROM lithology clay, silt, sand, peat depositional environment estuarine archaeological activity excavation (ARCHIS: p4257) executed by Archeologisch Onderzoek Leiden BV (ARCHOL) Introduction and aim Soil samples for magnetic susceptibility measurements were taken from Roman Period features during this excavation. This investigation was undertaken in order to clarify the results of the magnetometer survey in the central part of Harnaschpolder (see 8 Harnaschpolder). Could the lack of magnetic response of the archaeological features be explained by the magnetic susceptibility of the fills of archaeological features and the contrast with the matrix they are embedded in? Methodology Samples were taken from the fills of archaeological features and from the undisturbed matrix, three topsoil samples were also obtained from the section of the excavation. Results The results of the magnetic susceptibility measurements are displayed in Table 14. The magnetic susceptibility of the fill of the archaeological features and the undisturbed matrix that they are embedded in all fall within the same range of magnetic susceptibility, i.e. 7.61 to 11.23 x 10-8 m3/kg. The values for the topsoil samples are much higher. There is no consistent contrast, i.e. either consistently higher or consistently lower, between the features and the undisturbed matrix, but the largest contrast that was measured in these soil samples is 3.62 x 10-8 m3/kg, a contrast that is very weak with respect to the sensitivity of the magnetometer that was used during these studies. Furthermore, it is likely that variations in the much higher susceptibility topsoil are larger than the variations that are caused by the presence of archaeological features in the subsoil. Table 14 Magnetic susceptibility of the soil samples from the excavation in Harnaschpolder north. sample interpretation magnetic susceptibility

x 10-8 m3/kg 1 topsoil 32.33 2 topsoil 58.85 3 topsoil 50.80 4 undisturbed 7.61 5 undisturbed 11.09 6 undisturbed 7.92 7 ditch 9.98 8 ditch 10.40 9 ditch 7.73 10 ditch 10.04 11 main ditch 9.68 12 main ditch 11.23 13 posthole 7.93

140

Harnaschpolder south HP03S municipality Midden-Delfland central coordinates 81590, 448270 type of archaeological site settlement period ROM lithology clay, silt, sand, peat depositional environment estuarine archaeological activity excavation (ARCHIS: 3690) executed by ADC Archeoprojecten Introduction and aim See Harnaschpolder north. Methodology Magnetic susceptibility samples were taken from three Roman Period ditches. The samples that were taken from ditch G3 (Fig. 12) were investigated further by means of thermomagnetic measurements and IRM (Appendix II). Results The results of the magnetic susceptibility measurements are displayed in Table 15. The values for the magnetic susceptibility of the topsoil samples ranges from 13.81 to 32.15 x 10-8 m3/kg, values for the fill of archaeological features and the undisturbed C-horizon are all lower and fall within the narrow range of 9.60 to 12.89 x 10-8 m3/kg. There is no consistent magnetic susceptibility contrast between the archaeological and the undisturbed deposits. Based in the thermomagnetic measurements on the Curie balance (see Appendix II for data), it is not possible to distinguish between the topsoil, upper and lower fill and undisturbed C-horizon of ditch G3. The total magnetisation of all samples decreases slightly at rising temperatures, indicating the presence of some ferrimagnetic material in the samples, most likely magnetite or maghemite.

Figure 12 Ditch G3 from which samples for magnetic susceptibility, thermomagnetic and IRM measurements were taken.

141

Table 15 The magnetic susceptibility of the soil samples from the excavation in Harnaschpolder south. The depth at which the samples were taken from the sections is indicated between brackets and is measured from the base of the topsoil. sample interpretation magnetic susceptibility

x 10-8 m3/kg G1-1 topsoil over ditch (10 cm) 13.81 G1-2 upper fill ditch (40 cm) 11.70 G1-3 lower fill ditch (55 cm) 9.72 G1-4 topsoil not over ditch (10 cm) 15.55 G1-5 C next to ditch (45 cm) 10.51 G2-6 topsoil over ditch 32.15 G2-7 fill ditch 11.65 G2-8 C next to ditch 9.60 G3-9 topsoil over ditch 18.69 G3-10 upper fill 12.69 G3-11 lower fill 12.89 G3-12 C next to ditch 11.03 The interpretation of the IRM acquisition curves of the same samples is more discriminative (Table 16). Sample G3-9 (topsoil) and G3-10 (upper fill ditch) both contain magnetite and goethite in a similar ratio, the difference being that the topsoil sample contains a greater quantity of these iron oxides. These two samples are very different from G3-12 (undisturbed), containing oxidised magne-tite (probably maghemite) and hematite. The lower ditch fill, sample G3-11 contains magnetite and hematite. Although the samples of the fills of the archaeological features and the undisturbed material have very similar magnetic susceptibility values, they have distinctively different iron mineralogy. For a more detailed description see Chapter 5. Table 16 The results of the component analysis of the IRM data of the samples from ditch G3 in Harnasch-polder south. The value for the SIRM depends on the amount of magnetic material that is present in the sample, B1/2 is independent of the quantity of material and gives an indication of the type of magnetic material that is present in the sample (see Kruiver et al. 2001). DP is the dispersion factor, indicative for the width of the distri-bution. Low DP’s occur in crystalline material.

sam

ple

mag

netic

su

scep

tibili

ty

x10-8

m3 /k

g

mas

s x

10-3

kg

desc

ript

ion

com

pone

nt B1/2

mT SIRM A/m

SIR

M x

10

-3 A

m2 /k

g SIRM/� A/m

DP interpretation

cont

ribut

ion

to s

igna

l % Curie

balance

9 18.69 0.48 topsoil 1 38.9 0.95 19.7917 105894.418 0.39 magnetite 91 paramagnetic

2 1995.3 0.089 1.8542 9920.63492 0.43 goethite 9 - 10 12.69 0.52 upper

fill 1 40.7 0.218 4.1923 33036.3096 0.44 magnetite 90 para-

magnetic 2 1258.9 0.023 0.4423 3485.48221 0.45 goethite 10 - 11 12.89 0.49 lower

fill 1 43.7 0.119 2.4286 18840.7403 0.45 magnetite 94 para-

magnetic 2 501.2 0.008 0.1633 1266.60439 0.4 hematite 6 12 11.03 0.53 undis-

turbed 1 50.1 0.139 2.6226 23777.3482 0.44 maghemite 89 para-

magnetic 2 794.3 0.017 0.3208 2908.02101 0.45 hematite 11 -

Harnaschpolder east HP03E municipality Midden-Delfland central coordinates 81600, 447600 type of archaeological site off-site period ROM lithology clay, silt, sand, peat depositional environment estuarine archaeological activity test trenches (ARCHIS: 3927) executed by Gemeente Schipluiden

142

Introduction and aim The magnetometer survey of Harnaschpolder east, just to the south of the main Harnaschpolder survey (see 8 Harnaschpolder), was conducted before the excavation of a number of test trenches in which the presence of traces of a Roman Period parceling system was anticipated. These archaeo-logical features were expected to be similar to those in the main Harnaschpolder site, just to the north of this site, where the magnetometer survey was conducted after the excavation of test trenches. The aim of this survey was to directly link the magnetic anomalies from the magnetometer survey with the archaeological features that were excavated. Methodology The magnetometer survey covered three different fields. In two of the surveyed areas a test trench was excavated after the survey. Results There are very few magnetic anomalies in the magnetometer data (Fig. 13). Except from some scat-tered magnetic noise throughout the area as a whole, there is one linear magnetic anomaly with a positive and a negative component in the northern grid. No excavations have taken place here that could clarify the nature of the anomaly, but the intensity and the straightness of the anomaly suggest that it is likely that a pipe or a cable is buried here. Alternatively, a natural or man made feature with a fill that contains ferrimagnetic iron sulphides could be causing this type of anomaly. Two archaeological features were identified in the excavation of the test trenches that overlapped with the magnetometer survey. These two ditches are most likely to date from the Roman Period. In the results of the magnetometer survey there are no anomalies that correspond to the ditches that were excavated. There is insufficient magnetic contrast between the fill of the ditches and the matrix that they are embedded in to cause an anomaly which is detectable with the magnetometer that was used.

Figure 13 Results of the magnetometer survey in Harnaschpolder east. The location of the excavation trenches is indicated with a solid grey line, the Roman Period ditches with a thick grey line.

143

10 Heeten Heeten Hordelman HH03, HH04 municipality Raalte central coordinates 215900, 482600 type of archaeological site settlement, industrial (iron production) period ROM lithology sand depositional environment plaggensoil over aeolian archaeological activity excavation (ARCHIS: 3725) executed by ADC Archeoprojecten Introduction and aim In test trenches on the location Heeten Hordelman oost, traces of late Roman habitation were un-covered, an earlier excavation in the adjoining area showed evidence that this area hosted industrial iron production from the 3rd to the 5th century AD. A magnetometer survey was conducted in Heeten Hordelman oost prior to a full scale excavation, in order to assess if magnetic methods could be used to investigate the internal structure of metal working sites in a non-destructive way, and if these methods could be used to map previously unknown metal working sites in the eastern part of The Netherlands. Special attention was given to magnetic prospection on plaggensoils and the conse-quences of the occurrence of red sand for magnetometer surveys (see § 6.1). Methodology The magnetometer survey was carried out on a resolution of 1 x 0.25 meter in the central area, and a resolution of 0.5 x 0.5 meter in the smaller western an southern areas. Initially, soil samples for magnetic susceptibility measurements were collected by means of coring (Appendix III). During the excavation, the southern row of postholes of building 4 was sampled, and a series of samples was taken from the plaggensoil. One of the patches of red sand was investigated by taking magnetic susceptibility samples of the top of the C-horizon in a 1 x 1 meter grid. Results The results of the magnetometer surveys (top) and the archaeological excavations (bottom) are displayed in Figure 14. It is clear that the multitude of archaeological features that was revealed during the excavation is not reflected in the magnetometer plot. On the edges of the surveyed area, the interference of fences and buildings on the field edges is prominent. A few linear anomalies in the magnetometer data proved during excavation to be modern intrusions. The archaeological features that were detected in the magnetometer survey were metal working furnaces (stars) and a number of rubbish pits (arrows) which are likely to have been filled with burned material. It is remarkable that the fill of the other negative features does not have sufficient magnetic contrast to cause a magnetic anomaly. The magnetic susceptibility measurements show that the fill of the postholes of building 4 (Fig. 14 and Table 17) has a very low magnetic susceptibility when compared to the other samples. In fact, the magnetic susceptibility of the layers that the features have been cut into, the B- and the BC-horizon, is much higher, and the archaeological features could be expected to cause a negative magne-tic anomaly. Table 17 The minimum, maximum and mean magnetic susceptibility through the soil profile in Heeten Hordelman (HH03 samples) and of samples of the topsoil and of a series of postholes from the excavation (HH04 samples). magnetic susceptibility

(min) x 10-8 m3 /kg magnetic susceptibility (max) x 10-8 m3 /kg

magnetic susceptibility (median) x 10-8 m3 /kg

N

HH03 topsoil 18.88 32.55 21.33 12 HH04 topsoil 11.63 37.90 16.78 9 HH03 ?feature 14.28 16.86 15.57 2 HH04 posthole 4.86 20.22 14.82 5 HH03 B 10.16 51.67 25.15 7 HH03 BC 39.98 108.51 60.26 4 HH03 C 4.48 19.57 9.07 7

144

The reason that this negative contrast of the archaeological features can not be seen in the results of the magnetometer survey is possibly that the variability of magnetic susceptibility of both the topsoil (plaggensoil) and the subsoil (mainly the B- and BC-horizon) is greater than the magnetic suscepti-bility contrasts between the archaeological features and the matrix that they are embedded in. Investigations into the magnetic susceptibility of red sand showed that patches of red sand cause a discreet magnetic anomaly that resembles the anomaly that would be caused by an archaeological feature (see § 6.1).

Figure 14 The results of the magnetometer survey in Heeten Hordelman oost (top) and the archaeological exca-vation (bottom). Magnetometer data has been displayed in three ranges; western part: -3 to 4 nT, central part: -3 to 3 nT, and the southern part: -3 to 15 nT. The location of the furnaces is indicated with stars, the location of the rubbish pits that were detected in the magnetometer survey with arrows. The area that was surveyed is hatched (bottom).

145

The modeling of the magnetic behavior of the pit furnaces that are common in the Roman Period and early Middle Ages in the eastern part of The Netherlands showed that when surveying with a fluxgate gradiometer, a furnace can be detected down to a depth of 2 meters (see § 7.4). The grid spacing that is suitable to detect all the furnaces within this two meter depth bracket is 1 x 0.25 meter or 0.5 x 0.5 meter. It is likely that a set of unexcavated furnaces was detected in the southern area of the Heeten Hordelman survey. Here, a cluster of strongly magnetic anomalies probably represents a number of furnaces that is part of the group of furnaces that was excavated to the direct north. 11 Kolhorn Kolhorn - Waardpolder KH05 municipality Anna-Paulowna central coordinates 122382, 536836 type of archaeological site road, industrial (peat extraction) period LMED lithology silt, clay, peat depositional environment estuarine and peat formation archaeological activity excavation (ARCHIS: o32902) executed by Rijksdienst voor het Oudheidkundig Bodemonderzoek (ROB) Introduction and aim Between the 8th and the 14th century AD the peat area in and around De Waardpolder, north of Kolhorn, was inundated with sea water. After these inundations large scale peat extraction took place in the area, the traces of which are still visible in the landscape today, both on the ground and on aerial photographs. One of the features that is associated with the peat extraction in this area is the ‘road of Paludanus’, an almost 5 km long linear feature that was first recognized by Rutger Paludanus in 1776 (Van Geel & Borger 2002). He interpreted it as being a road on a dyke. The feature is still visible from the air, but due to ploughing, less so on the ground. Archaeological excavations in 1995 have shown that the road is in fact the last strip of unexcavated peat in the area. It is quite likely that it was deliberately left unexcavated so it could be used as a road. The aim of the survey was to investigate the magnetic and electrical contrast of the road (unexcavated peat) with the clastic matrix surrounding it. Methodology The location of the fieldwork was chosen with an aerial photograph that clearly showed the road of Paludanus as a soil mark (Topografische Dienst 1990). The area that was magnetically investigated measures 40 x 80 meter and was planned to be perpendicular to the road. Apart from the magneto-meter survey, a partly overlapping electrical resistance survey was conducted in an area of 20 x 80 meter. It was carried out with a TRCIA resistance meter with a spatial resolution of 1 x 1 meter and an instrument resolution of 0.1 Ohm. The probe distance between the mobile probes was 1 meter, and the depth of investigation can be assumed to be approximately 1 meter. Results Two magnetic anomalies (Fig. 15) could be mapped, a positive anomaly on the northern edge of the surveyed area and one in the southwest corner. The northern anomaly is related to a ditch that was partly filled in but could still be seen as a depression on the surface. The location of the linear anomaly in the southwest corner corresponds to that of the road of Paludanus, but the orientation of the anomaly is NNW-SSE, rather than the NW-SE orientation of the crop mark on the aerial photo. It is likely that the magnetic anomaly is caused by a feature that is related to the road, but not by the road itself. Two areas with high resistance values appear to be anomalous to the rest of the area, a linear anomaly on the north side, and a zone of high values in the middle of the surveyed area. The latter group of anomalies has a southwest-northeast direction and consists of strips of higher and lower resistance. Overall, the width of the zone of high resistance measures approximately 15 meters.

146

The location, width and direction of the anomalies correspond to the soil mark that is visible on the aerial photograph and represents the road of Paludanus. The northern anomaly corresponds to the ditch that was also mapped in the magnetometer survey. A number of cores was carried out in order to map the area. A typical profile consists from top to bottom of a patchy dark layer of approximately 40 cm of mixed materials (silt, clay and sand) with occasional (layers of) shell, followed by a 20 cm layer of peat. Under the peat is a sequence of silty, sandy and clayey light grey estuarine deposits, the reduction zone starts at 150 cm. There are no boreholes positioned over the geophysical anomalies. The results of the coring suggest that the remnant of the layer of peat is in the oxidized part of the soil matrix. This can explain the high resistance anomaly, it may be caused by pockets of air in the peat that makes up the road rather than a low resistance anomaly that would be expected to be caused by waterlogged peat. The fact that the road does not produce a negative magnetic anomaly as would be expected (see 21 Smokkelhoek) suggests that the contribution of diamagnetic water is of crucial importance for the strength of the magnetic anomaly.

Figure 15 The results of the geophysical survey in Kolhorn. Both datasets have been interpolated.

147

12 Limmen Limmen de Krocht LI03, LI04 municipality Castricum central coordinates 107600, 508600 type of archaeological site settlement, road period MED lithology sand depositional environment aeolian archaeological activity excavation (ARCHIS: m4658) executed by Amsterdams Archeologisch Centrum, Universiteit van Amsterdam (AAC-UvA) Introduction and aim In Limmen, a former scheduled archaeological site was excavated for scientific purposes. The site is a settlement and dates from the Middle Ages. It is located on a raised beach deposit which has been covered with peat and dune sand. Archaeological features are present in the latter deposit. During the first field season, a magnetometer survey could be conducted in an area that was excavated in the second field season. This allowed for direct comparison between the results of the magnetometer survey and the excavation. Moreover, archaeological features could be sampled during excavation for magnetic susceptibility measurements. Methodology A magnetometer survey was conducted on a resolution of 0.25 x 1.0 meter. Based on these results, the survey in the northeastern corner was duplicated on a resolution of 0.5 x 0.5 meter, using tape measures to locate each measurement. In both the 2003 and the 2004 season samples could be collected directly from the excavation trench. Topsoil and undisturbed material was collected from the sections, and one down-profile section was sampled in trench 7. In total, 63 samples could be collected from archaeological features during their excavation. These samples included ditch, well, pit and posthole fills. Results The results of the magnetometer surveys are displayed in Figure 16. The detailed survey in the north-eastern corner of the general survey has been boxed. A number of clear, mainly positive linear and circular anomalies can be seen, these are indicated in the interpretation diagram (Fig. 16, bottom). A hatch has been given to those anomalies that are caused by remanent, most likely metal, objects. The archaeological features that have been excavated are displayed as grey solids, and it can be seen that many of the magnetic anomalies correspond to archaeological features. Not all features cause a detectable signal, but only four of the detected magnetic anomalies do not have a clear archaeological cause. In order to investigate what kind of features were detected and why certain features could be mapped magnetically, whereas others could not, the area where excavation and magnetometer survey overlap is displayed in Figure 17. The interpretation diagram is annotated with the type of archaeological feature, and colours correspond to the three periods in which the site was inhabited, and to the post-habitation period. The most striking feature that was detected in the magnetometer survey is the couple of road ditches from the late period (1150-1250 AD). Two more ditches near the road ditches could be detected magnetically. Around and over the road five late period wells caused a magnetic anomaly. The anomalies that are caused by early and middle period features all represent wells too. In the post-habitation period pits have been dug for sand extraction or soil improvement, four of these have been mapped, as well as a late ditch. If compared to the multitude of features that was excavated, only a small number of features have caused a detectable anomaly. Except for the obvious road ditches, mainly wells were mapped during the magnetometer survey, but not all wells caused an anomaly. All the wells that were excavated in the area that was surveyed with the magnetometer are listed in Table 18. It appears from this table that there is no obvious correlation between the detecta-bility and for example period, depth of burial or construction method.

148

Figure 16 The results of the magnetometer survey at Limmen (top) with the detailed survey in the northeastern corner (box) and the interpretation diagram in black (bottom) as an overlay over the results of the excavation (grey). Magnetic anomalies that have been caused by metal objects have been hatched in the interpretation diagram. Most of the wells that could be mapped are constructed with plaggen, except for an early period barrel and box well. This construction material may have been beneficial to the enhancement of magnetic susceptibility, as the plaggen contain a high amount of organic material which is needed for magnetic susceptibility enhancement. The reason why wells, rather than other features, have a magnetic expression may be attributed to the organic nature of the construction material or the fill of the feature, maybe in combination with its volume.

149

Figure 17 A summary of the non-remanent magnetic anomalies that were identified and their relation to the archaeological record. Only the northern (excavated) half of the interpretation diagram in Figure 16 is shown. Anomalies that did not prove to be caused by an archaeological feature: grey line (no fill); early period (800-1000 AD): black line (no fill); middle period (1000-1150 AD): light grey fill; late period (1150-1250 AD): dark grey fill; post settlement: black fill. Table 18 List of wells that have been detected in the area where the magnetometer survey has been conducted. Phases are defined as early (800-1000 AD), middle (1000-1150 AD) and late (1150-1250 AD). The type of construction is listed together with the highest level at which the feature was encountered. The last column states whether or not the well was detected in the survey.

well phase construction top (N.A.P. in cm) detected 51 E plaggen + frame -2 n 33 E barrel 20 n 62 E plaggen + frame 25 n 47 E plaggen 30 n 54 E barrel 30 n 55 E plaggen + frame 24 n 35 E barrel + wickerwork unknown n 57 E barrel + plaggen 26 y 61 E box 15 y 46 E barrel 27 y 45 E plaggen + frame 34 y 43 M plaggen 7 n 66 M plaggen 7 n 63 M plaggen 10 n 42 M plaggen + frame 17 n 44 M plaggen 22 n 34 M barrel + plaggen + frame 30 n 58 M treetrunk 38 n 56 M barrel 48 n 32 M plaggen 10 y 65 M plaggen + frame 0 y 68 M plaggen 0 y 67 M plaggen 8 y 48 L wickerwork 11 n 60 L plaggen + frame 25 y 59 L plaggen + barrel 29 y 52 L plaggen 41 y 41 L plaggen 44 y 50 L plaggen 46 y

150

An investigation into the magnetic susceptibility of the wells led to the sampling of the top fill of well 37 (S4330 in trench 15) that was available for sampling (Fig. 18). The well is just inside the area that was surveyed, but unfortunately disturbance by metal fencing has overshadowed any possible magne-tic response of the well in the magnetometer data. The well is constructed with plaggen and dates from the late period. The magnetic susceptibility values were surprisingly low (4.82 and 5.22 x 10-8 m3/kg for the fill of the two samples that were collected from the fill), as were the values for the un-disturbed matrix (2.71, 3.05 and 4.03 x 10-8 m3/kg). The value for the sample from the iron stained soil falls into the same range (3.39 x 10-8 m3/kg). It is not clear if this well would have caused a detectable anomaly if the metal fence would not have influenced the measurements. It is clear, however, that the clastic secondary fill that was sampled here would not have caused it. One of the road side ditches that was clearly visible in the magnetometer data was sampled down-profile in a section. In Figure 19 a comparison is made between this dataset, and the down profile magnetic susceptibility of an undisturbed profile. The enhanced magnetic susceptibility of the ditch fill can clearly be seen, with values exceeding 200 x 10-8 m3/kg. The susceptibility of the other archaeological features is much lower, in Table 19 the magnetic susceptibility of all the features that were sampled, except for the well samples that were discussed above and the road ditch samples, are listed. The fills of the features are magnetically enhanced when compared to the undisturbed matrix, but the values are much lower than those measured on the samples of the road side ditch. This limited enhancement could, in combination with the volume of the features and their depth, add to the explanation why only a limited amount of features could be mapped during the magnetometer survey in Limmen. A more sensitive magnetometer could probably have produced better results. Why the ditch fills of the road side ditches have very high magnetic susceptibility values when compared to the other features is unknown. The reason why a number of wells have produced a detectable magnetic anomaly can possibly be attributed to a high magnetic susceptibility of the highly organic primary fill and organic construction material.

Figure 18 Sampling in progress on well 37 (S4330) in trench 15. The two left hand sample tubes are placed in the secondary fill of the well, the other tubes in undisturbed deposits. Iron staining in the soil can be seen in the one but last sample on the right hand side.

151

Table 19 The magnetic susceptibility of the samples from the excavation at Limmen in the 2003 and the 2004 season. Samples from the ditch fill of the road and the well have been excluded. All values x 10-8 m3/kg. N is the number of samples. maximum minumum median mean N topsoil 18.90 11.43 14.72 15.15 7 archaeological feature 84.13 4.84 13.39 17.15 42 undisturbed 8.15 4.68 6.06 6.31 9

-90.00

-80.00

-70.00

-60.00

-50.00

-40.00

-30.00

-20.00

-10.00

0.00

0.00 50.00 100.00 150.00 200.00 250.00

magnetic susceptibility x 10-8 m3/kg

dept

h in

cm

from

sur

face

Figure 19 A comparison between the down-profile magnetic susceptibility of an undisturbed section (the south facing section of trench 7) in grey dashed line and one of the road ditches (feature 4715) in solid black line. Note the high magnetic susceptibility values of the fill of the road ditch. 13 Meerssen Meerssen Onderste Herkenberg MH03 municipality Meerssen central coordinates 181350, 321500 type of archaeological site villa period ROM lithology silt depositional environment aeolian archaeological activity test trenches (ARCHIS: 13256) executed by Rijksdienst voor het Oudheidkundig Bodemonderzoek (ROB) Introduction and aim The remains of a partly excavated Roman villa were re-investigated by means of trial trenches by the Rijksdienst voor het Oudheidkundig Bodemonderzoek in order to outline the extend of the villa complex. A magnetometer survey was carried out to assess if the site could be delimited this way. A further aim was to investigate the magnetic susceptibility contrast between the limestone walls of the buildings and the loess matrix. On the eastern side of the villa complex a (Roman Period) chalk quarry was discovered in the 19th century, but is not visible at present. A magnetometer survey was conducted to investigate if this quarry would give a magnetic response.

152

Methodology The two areas were surveyed on a 0.5 x 1 meter grid spacing. After the magnetometer survey, the ex-cavation of the trial trenches started. The archaeological features were only uncovered, not excavated, resulting in the tentative interpretation of the features. Soil samples for magnetic susceptibility measurements could be collected from the top of the fill of the archaeological features that were present. Results The results of the magnetometer survey are displayed in Figure 20 (top) with an interpretation diagram (bottom). The anomalies that are likely to be caused by metal objects are indicated with a hatch in the interpretation diagram, and are not further discussed. The numbers that are used in the text refer to the numbers in the interpretation diagram. A group of anomalies (1, grey zone) with a strong positive and weakly negative component is located in the western part of the survey area. The strength and shape of the anomalies suggests that they are caused by remanent magnetism, rather than by induced magnetisation. Trench 2 partly overlaps the group of anomalies (Fig. 21), and the two anomalies that were clipped correspond to two pit like features. Their fill consists mainly of pieces of brick or tile, which could explain the remanent magnetic appearance of the anomalies. The results of the magnetic susceptibility measurements are displayed in Table 20. A sample of brick and rooftile was measured for its magnetic susceptibility, which is - as would be expected - very high. It has to be kept in mind, however, that the thermoremanent magnetization in the baked clay influences the magnetic susceptibility measurement. It is likely that the other anomalies in this group represent similar pit like features. To the east of 1, two linear negative magnetic anomalies can be seen (2 in Fig. 20). There is no trench in this location, so interpretation remains speculative, but could relate to limestone walls or foundation trenches. A limestone and a marl sample were collected from the foun-dation trenches of one of the buildings that belong to the Roman villa. Their magnetic susceptibility is very low (Table 20) when compared to their loess matrix, which means that walls or foundations constructed with this material would cause a negative magnetic anomaly. In the lower half of the magnetometer survey results some striping can be observed (dashed line in the interpretation diagram). These stripes occur along the slope of the edge of the loess plateau, and are likely to be caused by ploughing. In the western part of the strip that was surveyed, a positive magnetic anomaly (3 in Fig. 20) is caused by the presence of a pit like feature. Another similar feature that was observed in this trench did not cause a magnetic anomaly. Two of the features that were seen in the trench to the east of this do cause a positive magnetic anomaly, but the rest of the features in this group do not. On the eastern edge of the surveyed area a large positive anomaly was found to be the magnetic expression of a trench that was dug for a gas pipe. Two samples were taken from the trench fill (trench 9, S005 and S006), which have an increased magnetic susceptibility compared to the matrix that the trench has been cut into, the undisturbed loess. Table 20 The magnetic susceptibility of the samples from Meerssen onderste Herkenberg. sample location magnetic susceptibility

x 10-8 m3 /kg limestone trench 1, S012 -0.81 marl trench 1, S007 0.35 undisturbed loess 2 14.67 undisturbed loess 1 20.26 undisturbed loess 3 20.97 feature fill trench 1, S012 21.54 ditch fill trench 9, S006 29.20 ditch fill trench 9, S005 30.67 feature fill trench 1, S007 32.27 topsoil 3 32.78 feature fill trench 1, S011 34.11 feature fill trench 1, S010 35.59 topsoil 2 37.01 topsoil 1 39.48 brick / rooftile trench 1, S011 170.08

153

In Table 20 the samples have been sorted by increasing magnetic susceptibility. It is clear that all thearchaeological features that have been sampled have a higher magnetic susceptibility and the buildingmaterial limestone / marl has a lower magnetic susceptibility than the matrix of undisturbed loess.

Figure 20 The results of the magnetometer survey in Meerssen (top) and the interpretation diagram (bottom).Data has been interpolated. Magnetic anomalies that are caused by metal are hatched, as is a large modernfeature (5). Stripes in the data are indicated with a dashed line. Numbers refer to the text.

154

Two of the three topsoil samples have higher susceptibilities than the feature fills. Based on thesamples, the walls and the limestone rubble fill of the foundation trenches can be said to have a nega-tive magnetic contrast to the matrix, whereas the fill of the negative features has a positive contrast.The reason that not all the features that have been discovered during the trial trenching cause amagnetic anomaly can be contributed to four variables; the magnetic contrast, the size of the archaeo-logical feature, the depth of topsoil and the variation in topsoil magnetic susceptibility (see § 6.1).An investigation into the first three variables for the foundation trenches of the building that wereobserved in the western trial trench can be found in § 7.1. It is likely that the foundation trenches docause a magnetic anomaly, but it is probably outside the detection limits of the instrument that wasused. The application of a higher sensitivity magnetometer on this site could produce better results.A separate category is formed by the features that are filled with pieces of brick or tile, these causemagnetic anomalies with a positive and a negative component. This magnetic contrast is stronger thanthe induced magnetisation in the features on this site.

Figure 21 The interpretation diagram of the magnetometer survey (grey) and the results of the trial trenching(black).

14 Meteren

Meteren Hondsgemet ME03municipality Geldermalsencentral coordinates 149950, 430730type of archaeological site settlementperiod IA, ROMlithology claydepositional environment fluvialarchaeological activity test trenches (ARCHIS: 13889)executed by Archeologisch Instituut Vrije Universiteit (AIVU)

155

Introduction and aim Prior to the development of an industrial area at Meteren Hondsgemet, a series of trial trenches was excavated. A magnetometer survey could be conducted before the excavation, so a direct comparison could be made between the results of the magnetometer survey and the excavation.

Figure 22 The results of the magnetometer survey at Meteren (top) and intepretation diagram (bottom). Data has been interpolated and high pass filtered.

156

Methodology A magnetometer survey of 100 x 100 meter was conducted on a resolution of 0.5 x 1.0 meter. The data was interpolated to 0.5 x 0.5 meter, and a high pass filter was applied to remove a broad trend. Results The results of the magnetometer survey and an interpretation diagram have been displayed in Figure 22. There is relatively much variation in the dataset. Three linear anomalies with positive and negative components, marked with 1 in the interpretation diagram, can probably be attributed to the presence of pipes or cables in the subsoil. Other magnetic anomalies that are caused by metal objects are indicated with hatched circles. Two areas with an enhanced magnetic susceptibility can be dis-tinguished; one in the north and one in the south of the surveyed area (dashed line and marked with 2 in Fig. 22). Smaller positive anomalies are represented by grey circles. Three negative linear anoma-lies (3 in the interpretation diagram) can be seen in the eastern part of the data. A comparison of the interpretation diagram with the excavation data is displayed in Figure 23. Features that were dated in the recent past have a grey fill, and are indicated with R. Two of the three negative anomalies (3 in the interpretation diagram) are obviously caused by modern ditches. The third may also be caused by a feature of recent origin. The location of the houses that were excavated (annotated with H) is not directly reflected in the magnetometer data, but two of the houses are located in one of the areas that have an enhanced magnetic susceptibility. If this enhancement is caused by the presence of the houses, the other areas with an observed enhanced magnetic suscepti-bility enhancement could also be a possible indication for the presence of houses in these locations. None of the other archaeological features appear to be reflected in the magnetometer data.

Figure 23 Interpretation diagram and the results of the excavation. R indicates a modern feature, H an exca-vated house. Numbers are copied from the interpretation diagram (Fig. 22).

157

15 Oostende (Belgium) Oostende Raversijde (Belgium) OR04 municipality Oostende central coordinates - type of archaeological site settlement period PMED lithology sand depositional environment marine and aeolian archaeological activity - executed by - Introduction and aim Investigations in order to map any remains that could belong to the drowned village of Raversijde have been carried out within the framework of a Planarch project around the city of Oostende. A magnetometer survey was conducted on the beach where traces of the drowned village were expected. Methodology The magnetometer survey was conducted during low tide and as the tide was coming in on a resolu-tion of 0.25 x 1.0 meter.

Figure 24 The results of the magnetometer survey on the beach of Oostende. A zero mean traverse routine has been applied and the data has been interpolated. The sea is located to the west of the survey, the retaining wall to the east.

158

Results The results of the magnetometer survey are displayed in Figure 24. To the west of the surveyed area is the sea, to the east a retaining wall. In the eastern part of the survey, close to the wall, a number of remanent magnetic anomalies can be seen in the data, these may be caused by remains of the shelling that took place here in the Second World War, or by other metal or remanent magnetic objects. In the western part of the survey, two curving negative magnetic anomalies can be seen, that are parallel to the surf during low tide, and in the sea during high tide. These magnetic anomalies are likely to be caused by geological variations in the subsoil, possibly by beach cusps, sinusoidal formations of coarse sediment material, that often occur in intertidal zones. There are no anomalies in the magnetometer data that could be associated with the drowned village of Raversijde. Neither is there any archaeological evidence for its location. It is possible that the village is not or no longer buried in the area that was investigated. Other possibilities include a deeper burial of the remains or a lack of magnetic contrast between the building material or the fill of archaeolo-gical features and the undisturbed matrix. 16 Ossenisse Ossenisse Drogendijk OS05 municipality Hulst central coordinates 59000, 377900 type of archaeological site ?group of tumuli period ?BA, ?IA lithology clay, sand depositional environment estuarine archaeological activity - executed by - Introduction and aim Aerial photographs of the area around the Drogendijk showed clear crop mark features in a cereal crop in the summer of 2005. The nature of the underlying archaeological features was unknown, but morphologically resembled a group of tumuli. This type of feature, however, is unknown in this part of The Netherlands. Two areas were selected for a magnetometer survey. The aim of the survey was to obtain information about the nature and the dating of the features that were causing the crop marks. Moreover, it was tested if these features, with obvious different physical properties than the sur-rounding matrix, would also cause magnetic anomalies. Methodology Both the northern and the southern area were surveyed with a fluxgate magnetometer on a resolution of 0.25 x 1.0 meter. Figure 25 is the aerial photograph of the northern area, it was not possible to rectify the aerial photograph and the location of the survey remains speculative. It was attempted to include one of the circular features in the survey. Figure 26 shows the southern area. This photo too was too oblique to rectify, and the location of the survey has been sketched onto the aerial photo. Results The results of the magnetometer surveys and their interpretation have been displayed in Figure 27. In the northern area, there is one clear linear anomaly (1 in Fig. 27). This anomaly could not be linked to any of the crop marks in the aerial photograph, but is likely to represent a ditch of unknown date. The other anomaly is a very faint circular anomaly (2 in Fig. 27), which probably corresponds to the circular structure that is visible on the aerial photograph, although this can not be confirmed without the rectification of the aerial photograph. In Figure 28 the data containing the circle has been processed separately by applying a high pass filter and by interpolating the resulting dataset. It is interesting to see that the circle now appears to have an opening on its northwestern side, but this interpretation remains speculative as the filter may have caused the anomaly to change shape.

159

Figure 25 Aerial photograph of the northern survey area, looking south. The photograph is too oblique to rectify. The estimated location of the survey has been indicated with a black line. Photo by Dr. Adrie de Kraker, IGBA/VU Amsterdam.

Figure 26 Aerial photograph of the southern survey area, looking west. The photograph is too oblique to rectify. The estimated location of the survey has been indicated with a black line. Photo by Dr. Adrie de Kraker. In the southern area there are three clear linear anomalies which have an east-west direction (3 in Fig. 27), and which can be recognized on the aerial photograph. Two overlapping circular anomalies are marked with 4 in Figure 27. Because of the difference in size it is unlikely that these represent the same feature as the circle in the northern area. The fact that the two circular features are overlapping makes their interpretation as funerary monuments unlikely.

160

A series of mainly positive magnetic anomalies (5 in Fig. 27) is visible in the southern part of the survey. Some of these features can be seen on the aerial photograph and belong to what appears to be a structured group of crop marks. These anomalies may represent pit like features. The only north – south oriented linear anomaly, marked with 6, can be interpreted. It can be recognized as a trackway on the 1832 map1 of the area. The magnetometer investigations have shown that the features that are causing the crop marks in most cases do also cause a magnetic anomaly. The results of the magnetometer survey, however, could not aid the interpretation or date the features that are visible on the aerial photograph.

Figure 27 The results of the magnetometer surveys in Ossenisse (left) and the interpretation diagram of the magnetic anomalies (right). Data has been interpolated. Range of magnetic variation 0 – 2 nT (top) and -2 – 4 nT (bottom). Ferrous responses have been indicated with a hatch in the interpretation diagram.

1 Kadastrale minuut, 1832. Source: www.dewoonomgeving.nl.

161

Figure 28 Grid with the circular feature in the northern survey. A high pass filter has been applied to the data and the data has been interpolated. 17 Poeldijk Poeldijk PD03 municipality Westland central coordinates 75745, 449496 type of archaeological site settlement, ?road period ROM lithology silt, clay, sand depositional environment marine and aeolian archaeological activity excavation (ARCHIS: 4740) executed by ADC Archeoprojecten Introduction and aim A magnetometer survey was conducted on a plot of land in Poeldijk within the framework of the project Romeinse wegen in Den Haag (Roman roads in Den Haag). The aim of the survey was to assess if the stretch of Roman road that was projected on this location could be mapped by means of magnetic methods. Methodology The projected Roman road was expected to run along the northwest side of the greenhouse (Fig. 29). The magnetometer survey was carried out as close to the greenhouse as was possible. Soil samples for magnetic susceptibility measurements were collected during hand augering, the location of the samples was not recorded. Results The results of the magnetometer survey are displayed in Figure 29. The influence of the metal of the construction of the greenhouse on the magnetometer data is clearly visible on the southeastern side of the surveyed area. A number of highly magnetic remanent anomalies can be seen throughout the data, these are probably caused by pieces of metal. A dashed line indicates the location of a very weak magnetic curvilinear anomaly. There are no indications in the magnetometer data for the presence of a Roman Road, and it was shown in later excavations that its location was further south than was expected, and hence outside the area that was investigated. The results of the magnetic susceptibility measurements can be found in Table 21. The highest magnetic susceptibility was measured in a sample in which iron sulphides were recognized during sampling. A very low magnetic susceptibility, on the other hand, was found for the sample of dune sand. The magnetic susceptibility of the archaeological layer and the archaeological features that were sampled is lower than the topsoil susceptibilities but higher than the magnetic susceptibility of the undisturbed C-horizon.

162

Figure 29 The results of the magnetometer survey in Poeldijk. Data has been interpolated. The location of the magnetic anomaly is indicated with a dashed line. The magnetic susceptibility measurements show that there is a magnetic contrast between the archaeological layer and the archaeological features and the undisturbed matrix that they are embedded in. The largest contrast that was measured in these samples is 7.1 x 10-8 m3/kg. With such a contrast, a feature of 0.5 x 0.5 x 0.5 meter that is buried at a depth of 0.25 meter would produce an anomaly of 3 nT, an anomaly of such a strength can be detected in a magnetometer survey with the instrument used. The fact that the magnetic expression of the archaeological features is not visible in the magnetometer data is probably caused by a deeper burial of the features in combination with a variation in the topsoil magnetic susceptibility. The curvilinear magnetic anomaly that could be seen in Figure 29 is likely to be caused by a feature, probably a creek, in which formation of magnetic iron sulphides has taken place. A similar magnetic response was seen in Smokkelhoek (see 21 Smokkelhoek). See Chapter 5 for a more detailed discus-sion on iron sulphides. Table 21 The magnetic susceptibility of the samples fromPoeldijk. sample interpretation magnetic susceptibility

x 10-8 m3/kg 1 archaeological layer 15.63 3 topsoil 14.81 4 dune sand 0.73 5 C-horizon with phosphates 8.08 6 top of creek with iron sulphides 24.79 7 C-horizon with phosphates 8.53 8 archaeological feature 11.29 9 archaeological layer 12.27 10 C-horizon 6.16 11 archaeological feature 9.73 12 topsoil 17.32

163

18 Polre Polre PO05 municipality Bergen op Zoom central coordinates 74720, 393420 type of archaeological site Settlement period PMED lithology sand, clay, peat depositional environment Fluvial archaeological activity fieldwalking (ARCHIS: m4592) executed by Gemeente Bergen op Zoom Introduction and aim Through a combination of fieldwalking and information from historical maps, a possible location for the drowned village of Polre was established. A magnetometer survey was conducted in order to map any structural remains that could have belonged to Polre. It is assumed that the church was a brick building with a slate roof. Methodology The fieldwork concentrated on the highest part of a levee, which is still visible in the landscape after subsequent transgressions. A magnetometer survey was conducted, and a number of cores were taken in order to aid with the interpretation of the features.

Figure 30 The results of the magnetometer survey in Polre.

164

Results The results of the magnetometer survey are displayed in Figure 30. A multitude of magnetic anoma-lies can be seen in the data, their outline and interpretation can be found in Figure 31. The location of these cores is added to the interpretation map, the details of the cores can be found in Appendix III. The dominating anomaly is a linear feature on the east side of the surveyed area. This feature repre-sents a field boundary that is present on the map of 1832, but is currently not visible on the ground. The results of cores 5 and 6 confirm this interpretation. Core 5 is located on a high point of the levee, and the undisturbed sandy deposits start at a depth of 40 cm. In core 6, a thick layer of archaeological deposits (50 - 140 cm), overlies a layer of sandy clay in which the presence of iron sulphides was noticed. Underneath this layer is a layer of peat. The levee appears to be dropping off in the direction of core 6, it is likely that the linear feature is a ditch or moat around the church that was constructed along the edge of the levee. The magnetic anomaly may be caused by the archaeological deposits if these have an enhanced magnetic susceptibility, but soil samples were not collected on this location. Another possibility is that it is not the fill of the archaeological feature, but the layer underneath that is displaying anomalous magnetic behavior. The preferential formation of iron sulphides in the deposit underneath the ditch may have caused a ‘shadow’ magnetic anomaly of the archaeological feature (see Chapter 5 for a detailed discussion on iron sulphide formation). On the south side of the surveyed area, a less straight and less strong magnetic anomaly branches off the field boundary that was discussed above. This feature too can be related to a field boundary on the map of 1832. In cores 4 and 7, thick archaeological deposits could be recognized, and an enhanced magnetic susceptibility of these probably caused the magnetic anomaly. The difference in appearance between the north-south field boundary and the east-west field boundary could possibly be explained by the presence and absence of iron sulphide in the deposits underneath the ditch fill.

Figure 31 Interpretation of the magnetometer data from Polre and the location of the cores. Numbers refer to coring numbers.

165

Three smaller linear features can be observed abutting the main ditch in the center of the area under investigation. A comparison of core 1 and 2 suggests that a 40 cm deep feature, probably a small ditch, is causing the anomaly. There may be a fourth similar anomaly to the east of the main ditch. In the center of the small plot that is defined by the three linear anomalies, three strongly magnetic, well defined anomalies were mapped. In core 9, evidence of burning was found as well as a piece of iron slag. It is likely that the anomalies are caused by three high temperature features, possibly iron smelting furnaces. On the north side of the survey, three areas with a brick like magnetic response were identified, one of these was confirmed to be a wall or a foundation in core 8. 19 Raalte Raalte De Zegge VI RA03 municipality Raalte central coordinates 216663, 487868 type of archaeological site settlement (IA, EMED), industrial (iron production) (ROM) period IA, ROM, EMED lithology sand depositional environment plaggensoil over aeolian archaeological activity excavation (ARCHIS: 4006) executed by ADC Archeoprojecten Introduction and aim An archaeological excavation was conducted prior to the development of a housing estate. Remains of late Roman Period iron production were uncovered, and DW Consulting was invited to scan the archaeological remains in the excavation trenches with a magnetometer, in order to assess if there was enough magnetic contrast to map the buried features that were related to the iron production. Methodology The archaeological features were scanned in the excavation trench, using a Bartington GRAD601 fluxgate gradiometer. The relative anomaly strength was recorded. Soil samples were collected from the west side of the excavation trench, here, there were patches of red sand in the predominantly yellow sand, both of which were sampled. On the western edge of the trench, a sample was taken from a deposit of black sand that contained a large amount of organic matter. Moreover, the un-disturbed C-horizon and the topsoil were sampled. Further samples were taken from the iron ore that was used for the production of iron, and from a charcoal ring (earth kiln, houtskoolmeiler) that was used to carbonize wood for the iron production process. Results The results of the magnetic scanning can be found in Table 22, the results of the magnetic suscepti-bility measurements in Table 23. The topsoil, undisturbed C-horizon and black (high in organic matter) sand samples have a low magnetic susceptibility. This agrees well with the observation of a limited background variation during the magnetic scanning. The magnetic susceptibility of the red sand is higher than the background, and the patches of red sand caused a detectable magnetic anomaly during the scanning. Table 22 The results of the magnetic scanning in Raalte De Zegge VI. Data collected by DW Consulting, Barneveld, The Netherlands. feature magnetic variation background -2 to 2 nT charcoal ring -1 to 1 nT red sand 2 to 40 nT furnace -50 to 3000 nT

166

Table 23 The magnetic susceptibility of the samples taken from Raalte De Zegge VI. sample description magnetic susceptibility

x 10-8 m3/kg 1 topsoil 8.53 2 charcoal ring 72.40 3 yellow sand, C-horizon 10.53 4 yellow sand, C-horizon 10.55 5 black sand 0.68 6 red sand 56.48 7 iron ore 44.78 As could be expected, the magnetic anomaly caused by the furnaces is very strong, it is likely that the anomaly consists of a component of induced magnetisation and of a component of remanent magne-tisation. The iron ore has a high magnetic susceptibility, but also a remanent magnetisation. In the magnetic scanning, the charcoal ring did not appear to cause a magnetic anomaly. The soil sample that was taken from the feature, on the other hand, has a high magnetic susceptibility. These two observations do not agree, and the sample that was taken from the charcoal ring is probably not representative for the whole feature. The magnetic anomaly that is produced by the furnaces can easily be mapped even if the furnaces are buried. The patches of red sand create anomalies that can, in strength and in shape, be compared to the anomalies that are usually caused by archaeological features. These preliminary conclusions were tested on the site of Heeten Hordelman (see 10 Heeten), where the red sand patches as well as the detectability of furnaces was further investigated. 20 Slabroek Slabroekse Heide SB05 municipality Uden central coordinates 169830, 412456 type of archaeological site group of tumuli (BA), urnfield (IA) period BA, IA lithology sand depositional environment aeolian archaeological activity excavation and consolidation (ARCHIS: p13410) executed by Archeologisch Onderzoek Leiden BV (ARCHOL) Introduction and aim An excavation in 1923 by the Rijksmuseum voor Oudheden in Leiden identified this group of features on the Slabroekse Heide as a group of Bronze Age tumuli with an Iron Age urnfield. In 2005, limited excavation and consolidation of the remaining tumuli took place after the area was cleared of trees. A magnetometer survey was carried out as part of a larger geophysical investigation of the tumuli (Abdulfattah 2006) that was aimed at the assessment of the application of geophysical methods for mapping tumuli on sandy soils. Methodology Two individual tumuli and a blank area were investigated with a magnetometer. The southwestern quart of the eastern mound that was surveyed (Fig. 32) had already been excavated before the survey. The tumulus is visible in the field as a 1.3 meter high hillock. A magnetometer survey was conducted in order to assess if any traces of a primary or secondary burial or of the ring ditch that surrounds the tumulus, could be mapped. The central survey was laid out on the location of a suspected tumulus that was identified on an aerial photograph. On the surface there are no traces of the tumulus. The western survey area is located in a ploughed field. The aim of this part of the survey was to map any archaeological features that could be related to a western continuation of the urnfield or group of tumuli. There were no indications on the surface to suggest this.

167

Figure 32 The results of the magnetometer survey on the Slabroekse Heide. The location of the tumuli has been indicated with grey circles. Note that the plot is oriented towards the east, and that the displayed data range is diffferent for the western survey.

168

Results In the results of the magnetometer survey on the eastern mound no traces can be seen of burials or of a ring ditch, the only magnetic anomaly that possibly belongs to the tumulus is a remanent magnetic anomaly on its northern edge. This anomaly, however, is more likely to be related to the magnetic noise on the north side of the surveyed area, that is caused by the metalling of the track that runs here. In the central survey, no traces of a ring ditch surrounding the potential tumulus were mapped magnetically. Only one strong anomaly, probably a piece of metal, was detected during the survey. In the western survey a lot of magnetic noise can be seen that is probably caused by pieces of metal or brick that could also be seen on the surface of the ploughed field. No magnetic anomalies that could be an indication of the presence of any archaeological features were mapped during the survey. This is however no proof for the absence of archaeological features on this location. The ring ditch around the eastern tumulus, that was known to be present from its partial excavation, did not cause a detectable magnetic anomaly, and it is possible that the fill of the archaeological features on the site of Slabroekse Heide in general do not have sufficient magnetic contrast to be detected with the instrument that was used. 21 Smokkelhoek Smokkelhoek SH03 municipality Kapelle central coordinates 57380, 387530 type of archaeological site ?settlement (ROM), industrial (peat extraction) (MED) period ROM, MED lithology silt, sand, peat depositional environment estuarine and peat formation archaeological activity hand augering (ARCHIS: 8766) executed by RAAP Archeologisch Adviesbureau Introduction and aim A magnetometer survey was carried out after an archaeological prospection by hand augering found a possible Roman Period settlement, the remains of which were damaged by later peat extraction. The aim of the survey was to magnetically map the undisturbed peat in order to identify the locations where Roman Period remains had a potential to be preserved. Methodology A large scale magnetometer survey was carried out on a resolution of 1 x 0.25 meter. An area of 40 x 40 meter was selected for a detailed survey with a resolution of 0.5 x 0.25 meter. Based on the results of the surveys, a number of locations were selected for hand augering and soil sampling. All samples were freeze dried and their magnetic susceptibility measured. In cores 1 to 5 and 8 (Fig. 33) samples were taken to investigate the difference in magnetic susceptibility of peat, silt, sand and mixtures thereof. Cores 30 and 31 were sampled every 20 cm in order to get a magnetic susceptibility profile. Results In Figure 33 the results of the large scale magnetometer survey are displayed. The most obvious magnetic responses are the paired positive anomalies that dominate the magnetogram. Table 24 shows the magnetic susceptibility values from the samples that were taken from the cores over and next to these anomalies. In core 30 and 31 it can be seen that the high magnetic susceptibility values occur in certain bands, but through the whole reducing part of the soil section. The magnetic moment is probably carried by ferrimagnetic greigite as was the case in Harnaschpolder (see 8 Harnaschpolder) and Spalding (see 22 Spalding). It is likely that there is a magnetic remanence in the deposits but this has not been investigated, in a magnetometer survey induced and remanent magnetization can not be distinguished. Iron sulphide formation is a post depositional process, but morphologically its occur-rence appears to be related to sedimentary structures.

169

High magnetic susceptibilities were measured in silty clay and peat, but not in sandy deposits, which can be explained by the conditions under which iron sulphide formation can take place (Chapter 5). It is possible that the anomalies that can be seen in the data are negative anomalies (with a positive component on either side) caused by the weak magnetic moment of sandy creek deposits that cut through earlier layers in which iron sulphides could be formed.

Figure 33 The results of the magnetometer survey in Smokkelhoek with the location of the cores that are mentioned in the text. The box indicates the location of the detailed magnetometer survey (top). The inter-pretation diagram (bottom) has been overlayed with the historical map of 1832.

170

Table 24 The magnetic susceptibility of the soil samples from the cores in Smokkelhoek. core depth magnetic susceptibility

x 10-8 m3/kg interpretation oxidation /

reduction 1 310-315 37.20 silty clay R 1 350-360 2.98 sandy clay R 2 105-110 3.82 silty clay + OM + LP + LC + Fe O 2 190-200 4.18 silty clay, fill of peat tie R 2 210-215 -0.43 peat R 2 230-235 20.03 silty clay R 3 230-225 50.09 silty clay + OM, black layer R 4 130-135 6.17 silty clay + peat R 4 190-195 -0.30 peat R 4 210-215 99.01 silty clay R 5 110-115 3.02 silty clay + LP R 5 145-150 -0.80 peat R 8 215-220 13.21 peat R 30 0 6.68 silty clay O 30 20 6.83 silty clay O 30 40 5.67 silty clay + Fe O 30 60 4.13 sandy clay + Fe O 30 80 3.80 silty clay + Fe O 30 100 4.34 silty clay + Fe O 30 120 5.33 silty clay + Fe O 30 140 4.42 silty clay + Fe O 30 160 68.28 silty clay R 30 180 67.88 silty clay +OM +peat +black stains R 30 200 8.08 silty clay +OM +peat +black stains R 30 220 9.07 silty clay +OM +peat +black stains R 30 240 3.96 peat R 30 260 4.49 clayey peat R 30 270 18.12 silty clay R 31 0 6.64 silty clay O 31 20 6.56 silty clay O 31 40 6.75 silty clay O 31 60 5.86 silty clay + Fe O 31 80 7.04 silty clay + Fe O 31 100 6.93 silty clay + Fe O 31 110 7.86 silty clay + Fe O 31 120 0.87 peat, disturbed O 31 140 5.04 silty clay + OMs R 31 160 4.75 silty clay + OM R 31 180 -0.45 peat R 31 190 4.07 silty clay R 31 200 -0.46 peat R 31 220 58.93 silty clay R 31 240 177.33 silty clay R 31 260 104.73 silty clay R 31 280 177.28 silty clay R 31 300 35.20 silty clay R The second type of magnetic anomalies that can be seen in the data are patches of scattered magnetic noise, some of the responses are bipolar. Brick would give this type of response, and so would certain archaeological features. The overlay of the historical map of 1832 (Fig. 33) shows that the location of the magnetic anomalies coincides with the location of presently invisible and disused field bounda-ries. On the east side of the surveyed area a track can be seen both on the historical map (two parallel lines) and in the magnetometer data (positive anomaly). Magnetic responses that could be associated with former peat extraction could barely be seen in the coarsely spaced magnetometer survey. The location of the detailed magnetometer survey is indicated with a box in Figure 33. The data are displayed in Figure 34. In this survey, the peat ties and barrow ways can clearly be distinguished, the presence of peat is causing a negative magnetic anomaly. The interpretation of the data was confirmed with a series of cores along the line C-C’.

171

Figure 34 The results of the detailed magnetometer survey in Smokkelhoek (left) with interpretation (right) and the location of the coring transect that is referred to in the text. 22 Spalding (United Kingdom) Spalding Wygate, Lincolnshire, United Kingdom

SL04

municipality Spalding, Lincolnshire central coordinates National Grid Reference TF232 230 type of archaeological site industrial (salt production), settlement period ROM lithology silt, clay, peat depositional environment estuarine archaeological activity magnetometer survey, excavation executed by GSB Prospection Ltd., APS Archaeology, CgMs Midlands Introduction and aim In a magnetometer survey that was carried out by GSB Prospection Ltd., a series of strong magnetic anomalies was encountered that bore a morphological resemblance to the anomalies that were mapped in Harnaschpolder (see 8 Harnaschpolder) and Smokkelhoek (see 21 Smokkelhoek) within the frame-work of this study. It was expected that the magnetic variation was mainly caused by the presence of ferrimagnetic iron sulphides in the subsoil (Chapter 5). The aim of the hand auger survey was to asses where in the soil profile high magnetic susceptibility layers occurred, and, if possible, to identify the geological features that cause the magnetic anomalies. If the depositional environment in which the ferrimagnetic iron sulphides occur could be identified, the presence of these magnetic anomalies, which can hamper the interpretation of magnetometer data for archaeological purposes, could be anticipated upon. Methodology Nine boreholes in two transects were investigated by means of hand augering. Samples were taken from the retrieved sediment and were measured for their magnetic susceptibility without drying within 72 hours. The samples were freeze dried after the primary measurement and measured a second time, which resulted in a small decrease in magnetic susceptibility (see Table 4 in Chapter 4), being an indication that some magnetic changes have occurred in the process. Four samples were selected for Curie balance measurements, a sample from the topsoil and from the interface between the oxidizing and the reducing part of the soil section and two samples from the reducing part of the soil section (Fig. 36). Results In Figure 35 the location of the boreholes is superimposed on the results of the magnetometer survey that was carried out by GSB Prospection Ltd. for APS Archaeology /CgMs Midlands.

172

Figure 35 The results of the magnetometer survey at Spalding as collected by GSB Prospection Ltd. with the locations of the cores that were carried out within the framework of this study. The data is reproduced with permission of the owners: APS Archaeology and CgMs Midlands. The map is reproduced from Ordnance Survey Superplan Data © Crown Copyright 2004. All Rights reserved. Drawing reference number: 226486. A series of trial trenches showed that most of the magnetic anomalies that were mapped during the survey are not caused by archaeological features. A clear exception can be seen on the west side of the southern border, where an oval remanent magnetic anomaly proofed to be a Roman saltern. A profile of transect A is displayed in Figure 36. During the sampling it was clear that the black staining, indicative of the presence of iron sulphides, did not occur in the sandy creek deposits (grey solid in Figures 36 and 37), but rather in the low energy mud plain deposits. Iron sulphides only occur in a reducing environment, i.e. under the groundwater, although some black stains were noted just above the groundwater table. Down-core magnetic susceptibility measurements of core A6 and A7 are displayed next to the cores. In core A6, there is a sharp increase in magnetic susceptibility, which starts just above the groundwater table and continues below. In A7, the increase starts just underneath the groundwater table. In both cores, high magnetic susceptibility values occur in conjunction with black staining, mottling or layers, but the quantity or appearance of the iron sulphides could not directly be linked to the value for the magnetic susceptibility. The location of the Curie samples is displayed in Figure 36. The results of the Curie balance measure-ments can be found in Appendix II. Curie 1, a low magnetic susceptibility (21 x 10-8m3/kg) sample from the oxidizing part of the soil section was measured on the Curie balance. In the Curie plot a generally reversible decrease in total magnetization up to approximately 560 °C can be seen. This Curie temperature is slightly lower than the Tc of magnetite (Tc = 580 °C), which probably indicates of the presence of substituted magnetite in the sample. Sample Curie 2 has a high magnetic susceptibility (92 x 10-8m3/kg) and is derived from the interface of the reducing and oxidizing part of the soil column, from a section where both jarosite2 and black staining were present. The Curie plot of this sample shows an irreversible drop in magnetisation, which starts at room temperature and ends at approximately 320 °C. Although a fixed Curie tempe-rature for greigite cannot be established, this thermomagnetic behaviour could be typical for greigite. Neo-formation of a ferrimagnetic compound starts at approximately 410 °C, and has a peak around 500 °C after which the total magnetisation decreases irreversibly to approximately 580 °C, the Tc of magnetite.

2 Jarosite (KFe3(SO4)2 (OH)6) is an intermediate product of sulphide oxidation. It is easily visually identified as

yellow concretions.

173

Figure 36 Section of transect A. The figures next to core A6 and core A7 are magnetic susceptibility values x 10-8 m3/kg. Curie 1 to Curie 4 indicate the sample locations for the Curie balance measurements. The presence of sandy creek deposits is indicated with a grey solid fill. It is likely that this is the compound that was formed in the second part of the thermomagnetic run. On cooling no anomalies were recorded. The samples Curie 3 (magnetic susceptibility 281 x 10-8m3/kg) and Curie 4 (magnetic susceptibility 76 x 10-8m3/kg) were collected from the reducing part of the soil section. The Curie plots of these two samples resemble the plot of Curie 2, although the total magnetisation does not decrease as dramatically up to 320 °C as it does in the Curie 2 sample. It is likely that this decrease in the first part of the thermomagnetic run is caused by the break down of greigite. Neo-formation of a ferrimagnetic compound starts at 430°C for the Curie 4 sample, in the Curie 3 sample this is not as clear. Both plots show a drop in the total magnetisation at 580/590 °C, pin pointing the newly formed compound as magnetite. On cooling, the magnetisation increases from approximately 320 °C. In the Curie 3 plot there is a slight increase, whereas the Curie 4 sample reaches a much higher magnetisation, which is, at room temperature, comparable to the initial magnetisation of the sample. With a Tc of 320 °C, this increase can probably be attributed to the formation of pyrrhotite on cooling. The formation of iron compounds during the heating and cooling cycle is no indication of the original composition of the sample. The most important conclusion that can be drawn from the Curie balance measurements is that greigite is likely to be present in samples Curie 2 to Curie 4. In transect B (Fig. 37) it can be clearly seen that high magnetic susceptibility values do not occur in the sandy creek deposits, either under or above the groundwater table as is the case in core B2.

174

Figure 37 Section of transect B. The figures next to the cores are magnetic susceptibility values x 10-8 m3/kg. The presence of sandy creek deposits is indicated with a grey solid fill.

The same observations as in transect A were made regarding reducing circumstances and the presence of black stains, mottling and layers. The results of the coring and sampling can be used for a broad interpretation of the magnetometer data. Based on the Curie balance measurements, the main ferrimagnetic compound causing the magnetic anomalies could be identified as being greigite, although some magnetite was present in the topsoil sample. For further information refer to Chapter 5. The texture and depositional environment of the sediment appear to be crucial for the formation of iron sulphides. Sandy creek deposits do generally not host iron sulphides for a number of reasons (see Chapter 5 for details), but a mud plain depositional environment favors the formation iron sulphides. The magnetic anomalies are likely to be related to lateral changes in the depositional environment. Broadly speaking, creeks cause negative anomalies in the high magnetic susceptibility mud plains. Different phases in the estuarine landscape have different active creek systems. The magnetic anomalies that are measured at the surface are the sum of the magnetic expressions of these landscapes.

175

23 Stede Broec

Stede Broec Polder Het Valkje SB03 municipality Stede Broec central coordinates 144800, 524725 type of archaeological site ?settlement period BA lithology silt, sand, clay depositional environment estuarine archaeological activity none (ARCHIS: m1307) executed by -

Introduction and aim The Bronze Age settlement site Het Valkje is a scheduled archaeological monument. The site was first recognized as part of a larger Bronze Age landscape by aerial photography. The peat that has developed in the area after the Bronze Age habitation has been extracted in later periods. As a result, the Bronze Age landscape, and associated archaeological sites occur on or very near the present sur-face (see De Vries-Metz 1993 for more examples). During the magnetic investigations, flower bulbs had been planted at the core of the archaeological site, and while some samples for magnetic susceptibility measurements could be extracted, the magnetometer survey was relocated to the area to the east and northeast of Het Valkje. Table 25 The magnetic susceptibility of the samples from Stede Broec, sorted by coring location. The feature fills and the archaeological layer are not magnetically enhanced, except for the feature fill in D47.5. sample depth interpretation magnetic susceptibility

x 10-8 m3/kg C120A 25-30 topsoil with brick 11.61 C120A 45-50 ditch /feature with hard coal 6.08 C120A 65-70 ditch /feature 7.87 C120A 100-105 ditch /feature 4.67 C120A 155-160 undisturbed 3.65 C120A 165-170 undisturbed 9.53 C120A 170-175 undisturbed 9.11 C120A 175-180 undisturbed 11.98 C120A 180-185 undisturbed 9.33 C120A 185-190 undisturbed 8.19 C120A 190-195 undisturbed 3.83 D47.5 5-15 topsoil 10.16 D47.5 55-60 ditch /feature with hard coal 11.27 D47.5 60-65 ditch /feature 11.58 D47.5 95-100 ditch /feature 8.22 D47.5 100-105 ditch /feature 6.14 D47.5 130-140 ?undisturbed 3.05 D47.5 160-170 undisturbed 5.22 D100A 5-15 topsoil 13.49 D100A 70-75 ditch /feature with brick 5.61 D100A 100-105 ditch /feature 5.11 D100A 110-115 ditch /feature 3.83 D100A 150 undisturbed 3.31 D100A 215-220 undisturbed 6.02 50/25 5-15 topsoil 31.10 50/25 15-20 archaeological layer 7.95 50/25 30-40 archaeological layer 7.16 50/25 40-45 archaeological layer 5.32 50/25 70-80 undisturbed 4.94 50/25 110-120 undisturbed 3.50 Methodology Two strips of magnetometer data were collected on a resolution of 0.5 x 1.0 meter. The survey was located as close to the core of the archaeological site as was possible.

176

Samples for magnetic susceptibility measurements were collected from three cores in the survey area; C120, D47.5 and D100 (Fig. 38). Another set of samples was collected 250 meter to the south west of C120, at the core of the archaeological site ‘Het Valkje’, because earlier work had confirmed the presence of an archaeological layer in this location. Results The results of the magnetometer survey are displayed in Figure 38 (left) with an interpretation diagram on which the location of the cores is indicated (right). Only the cores from which samples are retrieved are displayed. Additional cores have shown that in the area south of C120 an archaeological layer is present. Only one induced linear magnetic anomaly can be recognized in the magnetometer data in this area. In C120 a fill of a feature, most likely a filled in ditch, was sampled, which contained modern debris like brick and hard coal. The magnetic susceptibility (Table 25) of the fill samples is not enhanced when compared to the undisturbed layers or the topsoil. It is likely, however, that the feature is associated with the body causing the magnetic anomaly, but it is possible that the enhanced layers are located under the feature. Two similar linear anomalies can be seen in the northern half of this strip of survey, but no intrusive investigations have been carried out there. The dashed line that is overcutting the survey is the representation of a set of overhead power cables, the influence of which can be seen as a zone of higher magnetic values. To the south of this zone an irregular, very magnetic feature was mapped, and a similar feature can be seen in the northern strip that was surveyed (grey solid fill in Fig. 38).

Figure 38 The results of the magnetometer survey in Stede Broec (left) and the interpretation diagram with the location of the cores (right). Data has been interpolated. The dashed line represents the orientation of the overhead power cables. In the interpretation diagram, anomalies that are interpreted as being caused by metal objects are hatched. Two anomalies with very strong remanent responses are displayed with a grey solid fill.

177

Core D47.5 was sampled here in order to aid the interpretation. In the core the magnetically enhanced fill of a feature could be recognized, which contained hard coal and probably has to be interpreted as a modern ditch fill. The magnetic anomalies have a remanent magnetic appearance, however, and cannot be caused by the magnetically enhanced fill that was encountered. The interpretation of these anomalies remains speculative. No further archaeological layers or fills could be recognized during the coring. In core D100 the magnetic susceptibility decreases down profile, and this part of the survey is magnetically quiet except for some magnetic spikes. Note the area of higher magnetic values under the influence of the metal construction of the greenhouse. The lack of any magnetic anomalies being caused by clear prehistoric features is confirmed by the samples from core 50/25. Here, a known Bronze Age layer was sampled which shows very little mag-netic susceptibility enhancement when compared to the subsoil, and no enhancement when compared to the topsoil. No Bronze Age features have been mapped during the magnetometer survey, and the features that could be mapped probably have to be dated as post Medieval. 24 Steenbergen Steenbergen Koevering SK03 municipality Steenbergen central coordinates 77370, 399310 type of archaeological site settlement period PMED lithology sand, clay depositional environment estuarine archaeological activity (ARCHIS: m15714 ) executed by - Introduction and aim The survey area at Steenbergen is located directly south of a farm with the name ‘Koevering’. This was the most likely location for the so-called drowned village of Koevering, based on the name of the farm in combination with historical research. The area is elevated with respect to its surroundings. A magnetometer survey was conducted within the framework of a multi-disciplinary project ‘Designing a drowned landscape’3 (Kluiving et al. 2007). The survey was conducted as a follow up from a hand auger survey, during which some pieces of brick were retrieved, in order to confirm the location of the village and to map any in situ settlement remains. Methodology The northeastern section of the field was surveyed on a 0.5 x 1 meter resolution. The dyke along which the current village of Koevering is built runs parallel to the eastern boundary of the survey area. Results The results of the magnetometer survey are displayed in Figure 39 (top) with an interpretation dia-gram (bottom). The data along the northeastern and southwestern boundary are affected by metal fencing. Striping in the data, which is due to ploughing, can be seen in the entire survey area (grey dashed line in the interpretation diagram). The most striking magnetic anomaly is located on the eastern side of the survey, parallel to the dyke. The patchy positive and negative anomaly is roughly linear in shape (dotted hatch). The results of the hand auger survey, conducted before the magnetometer survey, did not provide any information for a possible interpretation of the anomaly. The strength, consistency and shape resemble the anomalies that are caused by the presence of bodies of ferrimagnetic iron sulphides in for example Smokkelhoek (see 21 Smokkelhoek), Spalding (see 22 Spalding) and Harnaschpolder (see 8 Harnaschpolder); see Chapter 5 for more details.

3 ‘Een verdronken landschap vormgeven’ was funded by The Netherlands Architecture Fund. In the project,

geological, archaeological and historical data were integrated to form the point of origin for ideas for shaping the landscape in future developments. See Kluiving et al. (2007) for information on the results of the project.

178

In the central area a north-south oriented strip of magnetic noise can be seen (grey solid). The bipolar responses in this area may be an indication of the presence of brick. They are not likely to be caused by in situ brick walls or foundations, but rather by for example individual (pieces of) brick(s) that have been transported by water, like the material that was recorded during the hand auger survey. Inside the zone of magnetic noise an oval shaped negative magnetic anomaly can be seen, of which the interpretation is unknown. Although an archaeological origin for the magnetic anomalies in the central area can not be discarded entirely, they are more likely to be caused by geological phenomena, for example a small creek, possibly with brick material in its fill. If this is the case, it is possible that the brick has belonged to buildings in the drowned village of Koevering, but there are no indications for in situ brick walls in the magnetometer data that was collected, and the location of the drowned village can not be confirmed based on the magnetometer data.

Figure 39 The results of the magnetometer survey at Koevering (top) and the interpretation diagram (bottom). In the interpretation diagram a grey dashed line is used for striping due to ploughing, a grey solid for the area of magnetic noise and possible creek and a dotted hatch for the strongly magnetic anomalies that may have been caused by ferrimagnetic iron sulphides in the subsoil.

179

Historical research after the geophysical fieldwork showed that this location has also hosted a military battery of a later date than the drowned village of Koevering. The brick that was found in the hand auger survey and that may have been found in the magnetometer survey is possibly related to this building rather than to the drowned village, and the presence of this military building may also have contributed to the elevated nature of the field. 25 Swalmen Swalmen Kroppestraat AHN04 municipality Swalmen central coordinates 201995, 360120 type of archaeological site road period ROM lithology sand depositional environment fluvial archaeological activity (ARCHIS: m1383) executed by - Introduction and aim A stretch of Roman road near Swalmen is a scheduled archaeological monument and is visible on the surface as a strip of elevated ground. To the south of this easily visible part of the road is a stretch of road that is less easy to distinguish. This part has been mapped with LIDAR technology. To the north of the scheduled monument the course of the Roman road is not clear. A magnetometer survey was conducted on a part of the road which was clearly visible in order to investigate its magnetic expression. These results could possibly be used to map the course of the unknown stretch of Roman road with a magnetometer at a later stage. Methodology One 20 x 20 meter grid was surveyed on a resolution of 0.5 x 0.5 meter. Soils samples could not be collected. Results The results of the magnetometer survey and an interpretation diagram are displayed in Figure 40. Five circular magnetic anomalies, marked with 2 in the interpretation diagram, are located on the stretch of road. The two northern anomalies have a negative magnetic component, and may be caused by metal objects. The three southern anomalies are positive magnetic anomalies and may be caused by archaeological features or patches of material with a higher magnetic susceptibility. A linear negative magnetic anomaly is annotated with 1 in the interpretation diagram.

Figure 40 The results of the magnetometer survey in Swalmen (left) and the interpretation diagram (right).

180

The location of this anomaly coincides with the western edge of the Roman road. An interpretation of the anomaly as being caused by a road side ditches is plausible, but in excavations of other parts of this Roman road this type of ditch has not been observed. Because of the location and direction of the anomaly it is very likely that it reflects a feature that is associated with the Roman road. 26 Uitgeest Uitgeest UG04 municipality Uitgeest central coordinates 108853, 502763 type of archaeological site settlement period IA, ROM lithology sand, silt, peat depositional environment estuarine archaeological activity excavation (ARCHIS: p6662) executed by RAAP Archeologisch Adviesbureau Introduction and aim Archaeological excavations were carried out at Uitgeest prior to the widening of a field ditch. A limited number of traces from the late Iron Age / early Roman Period were uncovered. The excavation created the opportunity to sample the different units of this soil profile for magnetic susceptibility measurements. The measurements were conducted in order to assess the variation in magnetic sus-ceptibility in the oxidizing part of the soil in an estuarine / mudplain environment. Methodology The sampling locations are indicated in Figure 41, a description of the soil profile can be found in Table 26. Layer 1, at the bottom of the section, consists of silty sand and is followed by a layer of clayey peat (layer 2). These two layers represent an active depositional environment followed by much quieter circumstances in which peat could grow. Layer 4, a sequence of laminated sand and silt layers, is typical for a tidal mudplain environment.

Figure 41 The section at Uitgeest that was sampled for magnetic susceptibility. Numbers are sample numbers.

181

Table 26 The magnetic susceptibility of the samples from Uitgeest. sample interpretation magnetic susceptibility

x 10-8 m3/kg 1 undisturbed, silty sand 2.91 2 undisturbed, clayey peat 1.74 3 undisturbed, clayey silt 5.32 4 undisturbed, laminated sand / silt 2.56 5 topsoil, clayey sand 11.32 6 ditch fill 3.86 Results The magnetic susceptibility of the undisturbed material in the subsoil is very low, all the layers, from different depositional environments, have values that are lower than 6 x 10-8 m3/kg. The topsoil has a higher magnetic susceptibility. The ditch (sample 6) was cut into layers 2, 3 and 4. The magnetic susceptibility contrast between the archaeological feature and the undisturbed matrix is very small and inconsistent, i.e. both positive and negative. 27 Uitgeesterbroek Uitgeesterbroek UB03 municipality Uitgeest central coordinates 110620, 504040 type of archaeological site settlement period ROM lithology silt, sand, clay depositional environment estuarine archaeological activity (ARCHIS: m1317) executed by - Introduction and aim The scheduled archaeological monument in the Uitgeesterbroekpolder consists of a dwelling mound with habitation and three possible dwelling mounds, without clear habitation traces from the Roman Period. The former, most western mound measures 25 meter in diameter and has a height of a meter. The latter mounds are smaller and lower and lack archaeological evidence for habitation, they may have been used as higher ground for cattle. The mounds have been constructed on the elevated bedding of a dry creek system. The creeks are elevated when compared to the surroundings through a process of relief inversion. A magnetometer survey was carried out over two of the low mounds in order to map their magnetic response. If they would give a clear magnetic response the method could then be used to map a larger area for dwelling mounds that are less clearly visible. Methodology A magnetometer survey was carried out on a resolution of 1 x 0.25 meter. Six samples from archaeo-logical layers were collected for magnetic susceptibility measurements from the western mound, which was not surveyed (Fig. 42). Difficulties with the tie in may have resulted in positioning errors of up to 5 meters. The samples were taken by hand auger by the Rijksdienst voor het Oudheidkundig Bodemonderzoek. Results The results of the magnetic susceptibility measurements on the soil samples are displayed in Table 27. All the samples have a low magnetic susceptibility. No samples have been taken from the topsoil or the undisturbed matrix, which makes the contrast between the archaeological deposits and the other deposits unknown. The results of the magnetometer survey are displayed in Figure 42. There are no magnetic anomalies in the data that can be related to the mounds (circles in Fig. 42). It was observed in the field that the patch of magnetic noise in the northeastern corner of the survey was likely to be caused by pieces of building debris in that location.

182

Table 27 The magnetic susceptibility of the soil samples from Uitgeesterbroek. sample depth interpretation magnetic susceptibility

x 10-8 m3/kg 51E2 45-55 archaeological layer 9.30 51E1 25-35 archaeological layer 12.06 64F3 35-45 archaeological layer 7.80 62F2 35-45 archaeological layer 3.29 53E3 30-45 archaeological layer 8.06 60F1 35-50 archaeological layer 8.62 The magnetic detection of the dwelling mounds in the Uitgeesterbroekpolder appears not to be possible using the current methodology. The low magnetic susceptibility values from the archaeo-logical layers that were sampled agree with the lack of magnetic anomalies in the magnetometer data. Information about the possible cause for the lack of a magnetic contrast can be found in Chapter 5.

Figure 42 The results of the magnetometer survey in the Uitgeesterbroekpolder. Data has been interpolated. The approximate location of the mounds is displayed with circles. The western mound was not included in the mag-netometer survey, but samples for magnetic susceptibility measurements were collected from its archaeological layer.

183

28 Valkenisse Valkenisse VA03 municipality Reimerswaal central coordinates 65200, 380080 type of archaeological site settlement period LMED, PMED lithology sand, silt depositional environment estuarine archaeological activity plan of visible structures (ARCHIS: m13574) executed by Archeologische Werkgemeenschap Nederland (AWN) Introduction and aim The village of Valkenisse was first mentioned in the 13th century. Storm surges in the beginning of the 16th century damaged the buildings in the village, many of which were brick built. In 1682, after another heavy flood, the village was abandoned. In consequent years the remnants of Valkenisse were covered with a layer of sediments from the tidal river Schelde. In the 1990's, after the course of the river had been changed artificially, archaeological features that could be related to Valkenisse became visible again (Fig. 43). This created the opportunity to magnetically map a superficial drowned village site. The research was undertaken with the aim to record the magnetic responses of visible archaeological structures so that this information could be used for the interpretation of magnetometer data on invisible drowned village structures.

Figure 43 Magnetometer survey in progress in Valkenisse (left hand side). The two persons on the right are standing on the church foundations. In the background the section from which the undisturbed samples were taken can be seen. Methodology The magnetometer survey was carried out in two areas. The first grid measures 20 x 20 meter, and was placed over part of the visible church foundations (Fig. 43). The second grid was located to the south of the church in an area with superficial brick, stakes and wooden posts, and measured 20 x 30 meter. The resolution of the surveys is 0.5 x 0.5 meter. Because of the difficult walking conditions, measurements were individually located with tapes and triggered manually. Samples for magnetic susceptibility measurements were collected from the four layers of a furnace that was discovered during the survey, and of the undisturbed silt on which it was placed. Coal and slag from the furnace were also sampled. The fill of a posthole near the furnace could be sampled, and five pieces of brick were collected.

184

Figure 44 The location of the survey grid in relation to the church foundations (top) and the results of the magnetometer survey superimposed on the church foundations (bottom). Church plan adapted from Kuipers 1995.

Samples of the undisturbed matrix were taken near the church, and from the sediment that would have covered the village, these latter samples were taken from a section at the interface of the non-vegetated and the vegetated tidal land (visible in the background of Fig. 43). Results The results of the magnetometer survey in the first grid are displayed in Figure 44. The location of the survey was chosen such that a large part of the northern choir and part of the northern main wall were included, in addition to an area outside the church, with decreasing brick density. In the magnetometer data, there is a clear contrast between the area with brick and the area without. Due to the remanent magnetic nature of the bricks, they cause randomly oriented bipolar anomalies. The choir wall has produced a more consistent anomaly, with elongated positive and negative anomalies along the length of the wall. In this survey both information on the extend of the area in which brick is present, as well as some structural information could be obtained. The results of the second survey are displayed in Figure 45. The area in which the superficial brick occurs is clearly visible in the magnetometer data, but it is not possible to recognize any wall or foundation structures. It is likely that in this respect the results of a larger survey would have been easier to interpret.

185

In the southern half of the data two stronger anomalies can be recognized. During inspection in the field it could be assessed that the smaller anomaly is likely to be caused by a posthole with the post still in place. The fill of this posthole was sampled. The larger anomaly has been caused by a group of furnaces and metal working debris. The deposits in these furnaces too could be sampled. The results of the magnetic susceptibility measurements are displayed in Table 25. It is clear that the magnetic susceptibility of the furnace samples is very high, which fits well with the large magnetic anomaly that was observed in the magnetometer data. The susceptibility of the bricks that have been sampled is very high, but it has to be taken into account that the measurements have been influenced by the magnetic remanence in the objects. It is interesting to note that sample 2, from underneath the furnace, also has a very high magnetic susceptibility. Even though this layer is not part of the furnace, it has probably been exposed to high temperatures, enhancing the susceptibility of the material. The sample from the fill of the posthole has a lower magnetic susceptibility than the other archaeological samples, but there is a clear magnetic contrast between the undisturbed matrix and the fill of the posthole.

Figure 45 The results of the magnetometer survey south of the church (top) with the interpretation diagram (bottom). In the area with the dotted hatch superficial brick could be seen. A grey solid is used for those ano-malies that could be related to archaeological features in the field.

186

Table 25 The magnetic susceptibility of the samples from Valkenisse. Samples have been sorted from high to low susceptibility. sample description magnetic susceptibility

x 10-8 m3/kg 1 furnace, black layer 5665.60 4 furnace, black layer 4241.19 5 furnace, yellow layer 1169.34 3 furnace, red layer 983.47 6 furnace, slag 757.82 10 brick 213.06 7 furnace, coal 211.94 11 brick 203.20 2 under furnace, grey silt 202.25 13 brick 155.18 14 brick 100.39 9 brick 64.63 8 posthole 54.78 12 undisturbed, 60-90 cm 26.99 16 undisturbed, 60-90 cm 25.20 19 undisturbed, near church 22.74 15 undisturbed, 0-60 cm 22.49 17 undisturbed, near church 21.89 18 undisturbed 14.24 It remains unclear if it is indeed the magnetic susceptibility contrast that is causing the magnetic anomaly that could be seen in the results of the magnetometer survey. In this area boulders are often used as a solid foundation for wooden posts, and it is possible that a stone with a high magnetic sus-ceptibility or a magnetic remanence is present underneath the post, which would enhance the strength of the magnetic anomaly. 29 Wijk bij Duurstede Wijk bij Duurstede veilingterrein WD04 municipality Wijk bij Duurstede central coordinates 152000, 443000 type of archaeological site settlement period MED lithology clay depositional environment fluvial archaeological activity excavation (ARCHIS: 6025) executed by ADC Archeoprojecten and Archeologisch Centrum Vrije Universiteit –

Hendrik Brunsting Stichting (ACVU-HBS) Introduction and aim During a preliminary excavation on the location of the auction building of Wijk bij Duurstede, traces of Medieval habitation were uncovered. Soil samples for magnetic susceptibility measurements were taken in order to assess the magnetic contrast between the fill of the archaeological features and the undisturbed matrix that they are embedded in. Methodology The soil samples were taken during the excavation. Three topsoil samples, two samples of the undisturbed matrix and three samples of the fill of archaeological features were collected. Results The results of the magnetic susceptibility measurements are displayed in Table 26 The magnetic susceptibility of the two samples that were taken from the undisturbed C-horizon is low, between 9 and 13 x 10-8 m3/kg.

187

Table 26 The magnetic susceptibility of the samples from Wijk bij Duurstede. sample description magnetic susceptibility

x 10-8 m3/kg arch1 archaeological feature 83.23 arch2 archaeological feature 65.64 arch3 archaeological feature 76.11 C1 C-horizon

(undisturbed) 9.28

C2 C-horizon (undisturbed)

12.81

bv1 topsoil 240.48 bv2 topsoil 163.90 bv3 topsoil 129.73 There is a large magnetic contrast between the undisturbed matrix and the fill of the archaeological features, which have a magnetic susceptibility that is six to seven times higher. The topsoil samples have the highest magnetic susceptibility values. It is possible that the magnetic susceptibility of the topsoil samples, and to a lesser extend of the subsoil samples, is influenced by the semi-industrial activities that have taken place on the terrain of the auction building. 30 Zaltbommel Brakel Molenkampsweg ZB04 municipality Zaltbommel central coordinates 134870, 425135 type of archaeological site settlement period ROM lithology clay, silt depositional environment fluvial archaeological activity excavation (ARCHIS: 5590) executed by ADC Archeoprojecten Introduction and aim Traces of a Roman Period settlement were discovered and confirmed in an area of allotments at Brakel by means of hand augering and test trenching. During the planning of the full scale excavation, there was an opportunity to carry out magnetic measurements. Methodology A magnetometer survey was carried out on a resolution of 0.25 x 1.0 meter before the full scale exca-vation took place. During the excavation, soil samples for magnetic susceptibility measurements were collected from the undisturbed matrix, the fill of the negative archaeological features and the topsoil. Results The results of the magnetometer survey are displayed in Figure 46. A number of strongly positive and negative anomalies can be seen that are caused by metal objects. These objects are sheds, fence posts and fences that are erected in the allotments. The anomalies have been indicated with a hatch in the interpretation diagram. There is no indication in the data of any discreet anomalies that could be caused by archaeological features. Some trends can be seen in the eastern part of the survey, which are likely to be caused by geological changes in the subsoil, these are indicated with a dashed line in the interpretation diagram. The results of the archaeological excavation are displayed as an overlay over the interpretation diagram in Figure 47. The combination of the two datasets confirms that the archaeological features that have been excavated did not cause a detectable magnetic anomaly during the magnetometer survey. Moreover, the trends that could be recognized in the magnetometer data can not be related to the known archaeological record. The results of the magnetic susceptibility measurements have been digested in Table 27.

188

The fill of the archaeological features that were sampled generally has a slightly higher magnetic susceptibility than the subsoil. It can be assumed that some of the archaeological features have caused a weakly positive and others a weakly negative anomaly. These anomalies, however, have not been detected in the magnetometer survey.

Figure 46 The results of the magnetometer survey at Zaltbommel (top) and the interpretation diagram (bottom). Magnetic anomalies that have been caused by metal objects have been marked with a hatch, trends with a dashed line.

189

Figure 47 The results of the excavation at Zaltbommel (black) superimposed on the interpretation diagram of the magnetometer survey (grey). The detectablility depends on a number of variables; the strength of the magnetic susceptibility contrast, which is weak in this case, the volume of the archaeological feature and the depth of burial of the feature. It is likely that the combination of these variables at Brakel has been detrimental to the formation of detectable magnetic anomalies at the surface. The application of a magnetometer with a higher sensitivity may have yielded better results. Furthermore, if the variation of magnetic suscepti-bility in the topsoil layer is larger than the magnetic contrast between the archaeological feature and the matrix it is embedded in, the anomaly may be present at the surface but may not be recognized (see § 6.1). Table 27 The magnetic susceptibility of the samples from Zaltbommel.

minimum magnetic susceptibility x 10-8 m3/kg

maximum magnetic susceptibility x 10-8 m3/kg

median magnetic susceptibility x 10-8 m3/kg N

topsoil 16.70 27.97 17.15 5 archaeological feature 6.08 119.53 7.30 13 undisturbed 4.18 13.73 6.68 5 31 Zwaagdijk Oost Zwaagdijk Oost WH03 municipality Wervershoof central coordinates 138790, 524310 type of archaeological site settlement period BA lithology silt, sand depositional environment estuarine archaeological activity excavation (ARCHIS: 13876) executed by Archaeological Research en Consultancy (ARC)

190

Introduction and aim Archaeological excavations were conducted prior to the construction of an industrial estate at Zwaagdijk Oost. This site was selected for a magnetometer survey because it is representative for the area of eastern West-Friesland. Post Bronze Age marine influence has been very limited when compared to the western parts of Noord-Holland. Moreover, any peat that had developed after the Bronze Age has been excavated in later periods. As a result, the Bronze Age landscape, and asso-ciated archaeological sites occur on or very near the present surface (see De Vries-Metz 1993 for more examples). The site of Stede Broec (see 23 Stede Broec) has a similar setting. Methodology An area of 100 x 40 meter was surveyed by magnetometer on a 0.5 x 1 meter resolution. Because of the noisiness of the data, the dataset has been extrapolated to a 1 x 1 meter resolution. Excavation trenches were started after the survey. Samples were collected from a trench that was being excavated at the time of survey, not from the trench that overlaps the magnetometer survey. Results The results of the magnetometer survey (top left), the interpretation diagram (top right) and the excavation results with the interpretation diagram (bottom left) are displayed in Figure 48. Magnetic anomalies that are likely to be caused by metal objects are displayed with a hatch in the interpretation diagram. Two positive magnetic anomalies that could possibly be caused by archaeological features can be distinguished (1 and 2 in Fig. 48), but no excavations have taken place here. A magnetic trend in the data has been indicated with a dashed line. When compared to the results of the excavations, there appears to be no correlation between the trends and the excavation data, they are probably caused by geological variations or by magnetic susceptibility variations in the topsoil. The multitude of archaeological features that was uncovered during the excavation is not reflected in the results of the magnetometer survey. This observation corresponds well to the results of the magnetic suscepti-bility measurements of the soil samples (Table 28). The magnetic susceptibility of the samples is low. The topsoil samples and the modern ditch fill have relatively high magnetic susceptibilities, but even the magnetic susceptibility contrast between the samples with the highest (topsoil 8.80 x 10-8 m3/kg) and one of the lowest (ditch 2.79 x 10-8 m3/kg) magnetic susceptibility samples is minimal. In the limited set of samples that has been investigated, the range of magnetic susceptibility values of the undisturbed samples and the archaeological feature fill samples is comparable. There are no indica-tions either in the magnetometer data or in the magnetic susceptibility data that a magnetic contrast causing a detectable magnetic anomaly is present in the archaeological features on this site. Table 28 The magnetic susceptibility of the samples from Zwaagdijk Oost. sample interpretation magnetic susceptibility

x 10-8 m3/kg 1 topsoil 1 8.06 2 topsoil 2 7.85 3 topsoil 3 8.80 4 undisturbed 1 7.00

5 undisturbed 2 2.78

6 undisturbed 3 3.44 7 undisturbed 4 6.67 8 ditch (?BA) 4.83 9 ditch (ME) 6.07 10 ditch 2.79 11 posthole (?BA) 3.77 12 ditch (modern) 7.29

191

Figure 48 The results of the magnetometer survey at Zwaagdijk Oost (top left), the interpretation diagram (top right) and the results of the excavations superimposed on the interpretation diagram (bottom left). Magnetic anomalies that are likely to have been caused my metal objects are hatched.

192

193

Appendix II Laboratory data 1 Heating experiments Table 1A Broekpolder (BP02): Magnetic susceptibility (MS) measurements before heating. sample depth top

sample weight x 10-3 kg

weight container x 10-3 kg

weight sample x 10-3 kg

volume x 10-6 m3

MS ms11

weight corrected ms1

calibrated ms1 x 10-8 m3/kg

A80 5-20 5 12.39 2.81 9.58 9.7 2.1 10.96 14.52 A80 40-50 40 13.35 2.81 10.54 9 0.5 2.37 3.14 A80 85-90 85 15.9 2.81 13.09 10.5 0.6 2.29 3.04 A100 20-30 20 13.43 2.81 10.62 9.5 0.8 3.77 4.96 A140 25-35 25 16.09 2.81 13.28 11.5 1 3.77 4.96 A160 25-35 25 14.49 2.81 11.68 10 1.3 5.57 7.34 A180 15-45 15 14.07 2.81 11.26 9.5 1.1 4.88 6.47 B20 5-15 5 12.42 2.81 9.61 9.5 2.5 13.01 17.23 B40 30-40 30 14.69 2.81 11.88 9.7 1.1 4.63 6.13 1 ms1: measurement 1 Table 1A continued sample depth MS

ms22 weight corrected ms2


mean MS value x 10-8 m3/kg

A80 5-20 2.2 11.48 15.11 14.81 A80 40-50 0.6 2.85 3.75 3.44 A80 85-90 0.4 1.53 2.01 2.52 A100 20-30 0.9 4.24 5.58 5.27 A140 25-35 0.8 3.01 3.96 4.46 A160 25-35 1.5 6.42 8.45 7.89 A180 15-45 1.3 5.77 7.60 7.03 B20 5-15 2.3 11.97 15.75 16.49 B40 30-40 1.1 4.63 6.09 6.11 2 ms2: measurement 2 Table 1B Broekpolder (BP02): Magnetic susceptibility (MS) measurements after heating. sample depth weight

x 10-3 kg

weight container

volume x 10-6 m3

MS ms11




A80 5-20 11.22 2.805 8.5 58.5 347.59 465.520 8.415 A80 40-50 12.97 2.805 8.2 102.1 502.21 673.298 10.165 A80 85-90 15.57 2.805 10 23.2 90.87 120.539 12.765 A100 20-30 12.78 2.805 9 145.8 730.83 978.64 9.975 A140 25-35 15.21 2.805 10.5 194.8 785.17 1051.52 12.405 A160 25-35 13.16 2.805 8.7 137.2 662.48 886.97 10.355 A180 15-45 12.66 2.805 8 102.4 519.53 696.572 9.855 B20 5-15 11.4 2.805 8 64.3 374.05 501.078 8.595 B40 30-40 14.08 2.805 9.5 166.7 739.25 991.822 11.275 1 ms1: measurement 1

194

Table 1B continued sample depth MS


calibrated ms2

MS ms33



A80 5-20 56.4 335.12 447.96 57.6 342.25 457.42 A80 40-50 102.1 502.21 672.02 101.6 499.75 668.72 A80 85-90 23.3 91.27 120.98 22.3 87.35 115.47 A100 20-30 146.3 733.33 982.00 148.1 742.36 994.10 A140 25-35 191.7 772.67 1034.76 199 802.10 1074.22 A160 25-35 140.6 678.90 908.99 134.8 650.89 871.43 A180 15-45 97.7 495.69 663.27 99.8 506.34 677.55 B20 5-15 65 378.13 505.63 63 366.49 489.94 B40 30-40 167 740.58 991.64 164.9 731.26 979.29 2 ms2: measurement 2; 3 ms3: measurement 3 Table 1B continued sample depth mean MS value after

heating x 10-8 m3/kg mean MS value before heating x 10-8 m3/kg


A80 5-20 456.96 14.81 3.24 A80 40-50 671.34 3.44 0.51 A80 85-90 119.00 2.52 2.12 A100 20-30 984.91 5.27 0.54 A140 25-35 1053.50 4.46 0.42 A160 25-35 889.13 7.89 0.89 A180 15-45 679.13 7.03 1.04 B20 5-15 498.88 16.49 3.31 B40 30-40 987.58 6.11 0.62 Table 2A Harnaschpolder (HP02): Magnetic susceptibility (MS) measurements before heating. sample depth top

sample weight x 10-3 kg

weight container x 10-3 kg


volume x 10-6 m3

MS ms11



A120 25-45 25 13 2.81 10.20 10 2.1 10.30 13.64 A120 25-45 25 14 2.81 11.20 10 1.9 8.49 11.20 A140 25-35 25 14.96 2.81 12.16 10 1.7 6.99 9.22 A180 5-15 5 13.11 2.81 10.31 10 2 9.70 12.85 A180 35-40 35 13.4 2.81 10.60 9.5 1.9 8.97 11.88 A180 50-60 50 15.75 2.81 12.95 11 1.2 4.63 6.14 A180 50-60 50 14.53 2.81 11.73 10 1.2 5.12 6.76 A300 45-50 45 11.68 2.81 8.88 8.5 0.6 3.38 4.48 A40 30-35 30 9.07 2.81 6.27 5.5 0.9 7.18 9.47 A60 5-15 5 13.03 2.81 10.23 10 2.1 10.27 13.56 A60 30-40 30 11.92 2.81 9.12 8 1.1 6.03 7.95 A80 30-40 30 13.89 2.81 11.09 9.5 1.7 7.67 10.16 A80 30-40 30 13.65 2.81 10.85 9.5 1.4 6.45 8.52 B20 99-102 99 4.36 2.81 1.56 1.5 0 0.00 0.00 B40 40-45 40 15.09 2.81 12.29 11 1.5 6.11 8.06 B40 40-45 40 11.44 2.81 8.64 9.5 1 5.79 7.63 1 ms1: measurement 1

195

Table 2A continued sample depth MS



mean MS value x 10-8 m3/kg

A120 25-45 2.1 10.30 13.64 13.64 A120 25-45 2.1 9.38 12.34 11.77 A140 25-35 1.7 6.99 9.20 9.21 A180 5-15 2.2 10.67 14.14 13.50 A180 35-40 1.8 8.49 11.25 11.56 A180 50-60 1.4 5.41 7.16 6.65 A180 50-60 1.3 5.54 7.30 7.03 A300 45-50 0.6 3.38 4.48 4.48 A40 30-35 0.9 7.18 9.45 9.46 A60 5-15 2.3 11.25 14.80 14.18 A60 30-40 1.1 6.03 7.94 7.95 A80 30-40 1.7 7.67 10.16 10.16 A80 30-40 1.4 6.45 8.49 8.51 B20 99-102 0 0.00 0.00 0.00 B40 40-45 1.4 5.70 7.50 7.78 B40 40-45 1 5.79 7.62 7.63 2 ms2: measurement 2 Table 2B Harnaschpolder (HP02): Magnetic susceptibility (MS) measurements after heating. sample depth weight

x 10-3 kg

weight container

volume x 10-6 m3

MS ms11



weight sample 10-3 kg

A120 25-45 11.14 2.805 8.7 27.9 167.37 223.43 8.335 A120 25-45 12 2.805 9 29.1 158.24 211.17 9.195 A140 25-35 13.29 2.805 8.8 60 286.12 382.20 10.485 A180 5-15 10.49 2.805 8.1 43.9 285.62 382.31 7.685 A180 35-40 12 2.805 8.4 62.6 340.40 455.91 9.195 A180 50-60 14.77 2.805 10.5 20.9 87.34 115.91 11.965 A180 50-60 13.54 2.805 9 21.3 99.21 131.86 10.735 A300 45-50 9.77 2.805 7.2 87 624.55 837.66 6.965 A40 30-35 8.24 2.805 5.3 16 147.19 195.87 5.435 A60 5-15 11 2.805 8.5 34.6 211.10 282.19 8.195 A60 30-40 10.85 2.805 7.4 27.7 172.16 229.35 8.045 A80 30-40 12.4 2.805 8.5 37 192.81 257.61 9.595 A80 30-40 12.38 2.805 8.5 44.9 234.46 313.58 9.575 B20 99-102 4.42 2.805 1.2 59.7 1848.30 2481.76 1.615 B40 40-45 12.49 2.805 9.5 367 1894.68 2544.08 9.685 B40 40-45 9.47 2.805 6.5 227 1702.93 2282.41 6.665 1 ms1: measurement 1 Table 2B continued sample depth MS


calibrated ms2

MS ms33



A120 25-45 28 167.97 224.24 28.2 169.17 225.85 A120 25-45 28.6 155.52 207.52 28.9 157.15 209.71 A140 25-35 58.3 278.02 371.33 59.1 281.83 376.44 A180 5-15 42.5 276.51 370.07 43.4 282.37 377.94 A180 35-40 62.3 338.77 453.71 62.1 337.68 452.25 A180 50-60 20.6 86.08 114.23 20.3 84.83 112.55 A180 50-60 21.6 100.61 133.74 21.4 99.67 132.49 A300 45-50 85 610.19 818.37 84.9 609.48 817.41 A40 30-35 15.8 145.35 193.40 15.9 146.27 194.63 A60 5-15 35.1 214.15 286.29 36.3 221.48 296.13 A60 30-40 26.6 165.32 220.18 26 161.59 215.18 A80 30-40 36.3 189.16 252.71 36.7 191.25 255.51 A80 30-40 45.5 237.60 317.79 45.4 237.08 317.09 B20 99-102 60.8 1882.35 2527.52 65.6 2030.96 2727.17 B40 40-45 364 1879.19 2523.27 368 1899.85 2551.02 B40 40-45 235 1762.94 2362.91 246 1845.46 2473.58 2 ms2: measurement 2; 3 ms3: measurement 3

196

Table 2B continued sample depth mean MS value after

heating x 10-8 m3/kg mean MS value before heating x 10-8 m3/kg


A120 25-45 224.51 13.64 6.08 A120 25-45 209.46 11.77 5.62 A140 25-35 376.66 9.21 2.45 A180 5-15 376.77 13.50 3.58 A180 35-40 453.96 11.56 2.55 A180 50-60 114.23 6.65 5.82 A180 50-60 132.70 7.03 5.29 A300 45-50 824.48 4.48 0.54 A40 30-35 194.63 9.46 4.86 A60 5-15 288.21 14.18 4.92 A60 30-40 221.57 7.95 3.59 A80 30-40 255.28 10.16 3.98 A80 30-40 316.15 8.51 2.69 B20 99-102 2578.82 0.00 0.00 B40 40-45 2539.46 7.78 0.31 B40 40-45 2372.97 7.63 0.32 2 Curie Balance measurements 2.1 Introduction The characterization of a certain type of magnetic anomalies of a geological origin in the estuarine environment During a study into the application of magnetic methods for archaeological prospection in The Netherlands, magnetometer surveys in the western part of the Netherlands showed some unexpected results. Large scale, strongly magnetic anomalies with a creek like appearance dominated the magne-tic map of some of the archaeological sites under investigation. Low frequency magnetic susceptibi-lity measurements of soil samples obtained by hand augering, showed that the high magnetic suscepti-bility sediment was waterlogged, and the magnetic susceptibility decreased strongly upon oxidation. Black staining, associated with the presence of pyrite, occurred frequently in the high susceptibility layers or just above or below it. It is postulated that the magnetic anomalies are caused by the presence of bodies of iron sulphides in the subsoil. After the presentation of the preliminary data of the results of this project on the Archaeological Pros-pection conference in Krakow (2003), some of the other research groups and commercial companies working in archaeological prospection in the UK and Ireland, recognized the type of anomalies that were mapped in The Netherlands from their own surveys on estuarine deposits. No characterization of these anomalies has been made, however, and it is likely that often these natural anomalies have been interpreted as archaeological features. In order to understand the formation and preservation of these magnetic anomalies in the estuarine environment, an investigation of the iron mineralogy of some of the soil samples is carried out. In order to investigate whether it is the same phenomena causing the magnetic anomalies in the Netherlands as it is in the UK, some samples from Spalding, UK, are included in the research. The lack of magnetic contrast between archaeological features and the undisturbed matrix in the estuarine environment Magnetometry is a technique that is widely used to map archaeological features. In many instances the fill of for example ditch and pit features has a higher magnetic susceptibility than the undisturbed matrix. A number of processes can cause this magnetic contrast; • The feature is filled with a different material than it is cut into, the magnetic susceptibility of the

feature can be either higher or lower than the magnetic susceptibility of the undisturbed matrix. • The feature is filled with topsoil material, topsoil generally has a higher magnetic susceptibility

than the subsoil. • The feature is filled with settlement material which generally has a higher magnetic susceptibility

due to the magnetic material caused by burning.

197

The magnetometer surveys that have been carried out in the estuarine areas of The Netherlands seem to lack this sort of magnetic contrast. Measurements of the low frequency magnetic susceptibility of soil samples collected at excavations and from cores show that the magnetic susceptibility of the sediments is generally very low (approx. 5 to 10 x10-8 m3/kg), and the magnetic susceptibility of the fills of archaeological features seem to fall in the same range. Modern features (filled in ditches of the 20th century) dug in the same material have however obtained a magnetic contrast. All the sites that have been investigated in the estuarine area of The Netherlands have had marine transgressions after the abandonment of the archaeological site. It is proposed that these marine trans-gressions have changed the iron mineralogy of the soils and by doing so have deleted the magnetic contrast between the archaeological features and the undisturbed matrix. A magnetic investigation of the iron mineralogy of a number of soil samples is needed to investigate these processes further. 2.2 Sample selection The characterization of a certain type of magnetic anomalies of a geological origin in the estuarine environment The Netherlands Harnaschpolder Core PYR depth 5-15 mag sus 17 x 10-8 m3/kg topsoil O Core PYR depth 30-35 mag sus 11 x 10-8 m3/kg Duinkerke O Core PYR depth 372-374 mag sus 9 x 10-8 m3/kg Calais R Core PYR depth 374-376 mag sus 15 x 10-8 m3/kg Calais R Core PYR depth 376-378 mag sus 114 x 10-8 m3/kg Calais R Core PYR depth 378-380 mag sus 135 x 10-8 m3/kg Calais R Core PYR depth 395-400 mag sus 64 x 10-8 m3/kg Calais R Sample treatment: Samples were frozen immediately after sampling and freeze dried within two weeks, after freeze drying samples were kept in the freezer under silica gel. United Kingdom Spalding Core A6 depth 30 mag sus 21 x 10-8 m3/kg silt (topsoil) O Core A6 depth 170 mag sus 92 x 10-8 m3/kg silt with black stains / jarosite R Core A6 depth 300 mag sus 281 x 10-8 m3/kg silt with black stains R Core A6 depth 440 mag sus 76 x 10-8 m3/kg silt with black stains R Sample treatment: Samples were frozen immediately after sampling and freeze dried within three months, after freeze drying samples were kept in the freezer under silica gel. The lack of magnetic contrast between archaeological features and the undisturbed matrix in the estuarine environment The Netherlands Harnaschpolder-South G3-9 mag sus 18.69 x 10-8 m3/kg topsoil over ditch G3-10 mag sus 12.69 x 10-8 m3/kg upper fill G3-11 mag sus 12.89 x 10-8 m3/kg lower fill G3-12 mag sus 11.03 x 10-8 m3/kg C next to ditch

2.3 Methodology Soil samples were obtained either by hand-augering (Harnaschpolder and Spalding) or from the sections of archeological excavations (Harnaschpolder-South and Limmen) or from a freshly cut ditch (Broekpolder). Samples were stored in lidded plastic sample tubes, and were either freeze dried (Harnaschpolder and Spalding) or air dried (Broekpolder, Harnaschpolder-South and Limmen). The magnetic susceptibility of all samples has been measured using a AGICO KLY-2 susceptibility bridge in the Palaeomagnetic Laboratory ‘Fort Hoofddijk’ of the Universiteit van Utrecht. Thermomagnetic measurements were carried out with a modified horizontal-translation-type Curie Balance (Mullender et al. 1993) with a sensitivity of approximately 5 x 10-9 Am2 in the Palaeomag-netic Laboratory ‘Fort Hoofddijk’ of the Universiteit van Utrecht.

198

Samples were placed in a quartz glass sample holder and kept in place with quartz glass wool, and heated and cooled in air in 16 runs of 15-150, 150-50, 50-250, 250-150, 150-300, 300-200, 200-350, 350-250, 250-400, 400-300, 300-500, 500-400, 400-600, 600-500, 500-650, 650-15 °C, at a heating rate of 10 °C and a cooling rate of 15 °C per minute. The alternating field varied from 150-300 mT. 2.4 Results The characterization of a certain type of magnetic anomalies of a geological origin in the estuarine environment Netherlands Harnaschpolder

Figure 1 The results of the Curie balance measurement of sample core PYR, depth 5-15 cm, magnetic suscepti-bility 17 x 10-8 m3/kg. Low magnetic susceptibility – oxidizing (5-15 and 30-35 cm) Two samples from the oxidizing part of the soil section were measured on the Curie balance. The magnetic susceptibility of these samples was low at 17 and 11 x 10-8 m3/kg. In the Curie plots there is no evidence for the presence of a certain type of ferrimagnetic iron compound, the hyperbolic shape of the plot is caused by paramagnetic material, most likely the clastic material that forms the bulk of the samples. The total magnetization of both samples decreases after being heated to 650 °C, which indicates that there is some ferrimagnetic material present in the samples. Low magnetic susceptibility – reducing (372-374 cm and 374-376 cm) Two samples from the top of the reducing part of the soil section were submitted to a series of thermomagnetic runs. In the lower temperature part of the graphs there is an irreversible decrease of the total magnetisation up to a temperature of respectively 320 or 350 °C. This decrease is clearer in the 374-376 cm plot than it is in the 372-374 cm plot.

199

Figure 2 The results of the Curie balance measurement of sample core PYR, depth 30-35 cm, magnetic susceptibility 11 x 10-8 m3/kg. The thermomagnetic behaviour of greigite (Fe3S4) has been described by Dekkers et al. (2000), who state that greigite shows a typical decrease of magnetization between ~250 °C and ~350 °C, depending on the grain size of the material. A Curie temperature for greigite is difficult to assess because of the thermal decomposition of the compound that occurs at low temperatures. Both plots are dominated by the neo-formation of a ferrimagnetic compound above the temperature of 400 °C. The irreversible decrease of the total magnetization from respectively 480 and 500 °C up to 580 °C may identify the newly formed compound as magnetite (Tc = 580 °C), although apparent chemical chances may influence the total magnetization and apparent Curie temperature as well. Greigite that is heated under air will form magnetite starting from 400 °C in natural samples (Dekkers et al. 2000), but the oxidation of pyrite to magnetite starts at approximately 420 °C (Van Velzen & Zijderveld 1992). Given the low magnetic susceptibility of these two samples (respectively 9 and 15 x 10-8 m3/kg) only a very limited amount of greigite is expected to be present. The neoformation of magnetite may in this case mainly be caused by the oxidation of pyrite rather than by the oxidative alteration of greigite. High magnetic susceptibility – reducing (376-378 cm, 378-380 cm and 395-400 cm) Three samples with a high magnetic susceptibility (respectively 114, 135 and 64 x 10-8 m3/kg) from the lower reducing part of the soil section were thermomagnetically investigated. This series of Curie plots is dominated by the irreversible drop in magnetization at lower temperature ranges, from room temperature to 320 °C for the upper two samples (376-378 cm and 378-380 cm) and to 370 °C for the lowest sample (395-400 cm), the former may well indicate the presence of greigite in the sample, the latter temperature, on the other hand, is higher than temperatures that generally have been observed for synthetic and natural greigite samples, and may indicate the presence of (newly formed?) phyrrhotite (Tc = 420 °C) in this sample.

200

Figure 3 The results of the Curie balance measurement of sample core PYR, depth 372-374 cm, magnetic susceptibility 9 x 10-8 m3/kg.

Figure 4 The results of the Curie balance measurement of sample core PYR, depth 374-376 cm, magnetic susceptibility 15 x 10- 8m3/kg.

201

Figure 5 The results of the Curie balance measurement of sample core PYR, depth 376-378 cm, magnetic susceptibility 114 x 10-8m3/kg.

Figure 6 The results of the Curie balance measurement of sample core PYR, depth 374-376 cm, magnetic susceptibility 135 x 10-8 m3/kg.

202

Figure 7 The results of the Curie balance measurement of sample core PYR, depth 395-400 cm, magnetic susceptibility 64 x 10- 8 m3/kg. Neo-formation of a ferrimagnetic compound starts at approximately 410 °C. The total magnetization decreases irreversibly to approximately 600 °C to 620 °C, these temperatures suggest that maghemite (Tc ~ 600-675 °C) rather than magnetite (Tc = 580 °C) may be the final ferrimagnetic product of the thermomagnetic run. Because of the obvious chemical alterations in the sample during the thermomagnetic investigations, this remains highly speculative. The contributions of greigite and/or pyrite to the newly formed compound cannot be separated. High magnetic susceptibility values for these three samples suggest the presence of a considerable amount of greigite in the samples. United Kingdom: Spalding, Lincolnshire Low magnetic susceptibility – oxidizing (30 cm) One sample from the oxidizing part of the soil section was measured on the Curie balance. The magnetic susceptibility of this samples was low at 21 x 10-8 m3/kg. The Curie plot shows a mainly reversible decrease in total magnetization up to about 570 °C. This could indicate a small amount of substituted magnetite to be present in the sample. This substitution would cause the Curie temperature to be slightly lower than for pure magnetite. High magnetic susceptibility – oxidizing / reducing (170 cm) A sample with a high magnetic susceptibility (92 x 10-8 m3/kg) from the interface of the reducing and oxidizing environment (both jarosite and black staining present) was thermomagnetically investigated. This Curie plot resembles the high magnetic susceptibility plots of Harnaschpolder. An irreverible drop in magnetization starts at room temperature and ends at approximately 320 °C. This may indicate the presence of greigite in the sample, although a fixed Curie temperature for greigite cannot be established. Neo-formation of a ferrimagnetic compound starts at approximately 410 °C, and has a peak around 500 °C. After that the total magnetization decreases irreversibly to approximately 580 °C. This pro-cess is not reversible, and chemical changes obviously occur. It is likely, however, that in this second part of the thermomagnetic run magnetite was formed. On cooling no anomalies were recorded.

203

Figure 8 The results of the Curie balance measurement of sample core A6, depth 30 cm, magnetic susceptibility 21 x 10-8 m3/kg.

Figure 9 The results of the Curie balance measurement of sample core A6, depth 170 cm, magnetic suscepti-bility 92 x 10-8 m3/kg.

204



205

High magnetic susceptibility – reducing (300 and 440 cm) Two samples from reducing circumstances were selected for thermomagnetic measurements. The Curie plots of these two samples do resemble the 170 cm plot, although the total magnetization does not decrease as much up to 320 °C as it does in the former sample. Neo-formation of a ferrimagnetic compound starts at 430 °C for the 440 cm sample, in the 300 cm sample this is not as clear. Both plots show a drop in the total magnetization at 580/590 °C. On cooling the magnetisation increases down from about 350 °C. In the 300 cm plot there is a slight increase, whereas the 440 cm sample reaches a magnetization at room temperature that is comparable to the magnetization it started with. It is possible that the samples were not completely oxidized during the thermomagnetic run and that pyrrhotite (Tc = 320 °C) has formed at lower temperatures. The lack of magnetic contrast between archaeological features and the undisturbed matrix in the estuarine environment The Netherlands Harnaschpolder South

Figure 12 The results of the Curie balance measurement of sample G3-9, magnetic susceptibility 19 x 10-8

m3/kg. Four samples from an archaeological context were selected for thermomagnetic measurements. All samples show mainly paramagnetic behaviour during the thermomagnetic run. Although the indivi-dual components cannot be recognized, it can be observed that the total magnetization decreases. It is assumed that this indicates that a ferrimagnetic compound, most likely from the magnetite-maghe-mite series, has disappeared during the thermomagnetic run.

206


m3/kg.


m3/kg.

207


m3/kg. 3 IRM component analysis Broekpolder sample description component B1/2 mT SIRM (A/m) DP 1-1 topsoil 1 41.2 52500 0.29 2 120.2 16500 0.35 3 1995.3 7000 0.40 1-2 undisturbed 1 56.2 13700 0.46 2 707.9 1400 0.30 1-3 undisturbed 1 56.2 13200 0.40 2 707.9 2200 0.37 1-4 topsoil 1 42.7 88200 0.35 2 501.2 8500 0.47 1-5 feature 1 53.7 12400 0.42 2 501.2 850 0.30 1-6 feature 1 52.5 10600 0.38 2 446.7 1050 0.48 SEA modern feature 1 41.7 55500 0.38 2 631.0 3000 0.2 Harnaschpolder South sample description component B1/2 mT SIRM (A/m) DP G3-9 topsoil 1 38.9 95000 0.39 2 1995.3 8900 0.43 G3-10 undisturbed 1 40.7 21800 0.44 2 1258.9 2300 0.45 G3-11 undisturbed 1 43.7 11900 0.45 2 501.2 800 0.40 G3-12 topsoil 1 50.1 13900 0.44 2 794.3 1700 0.45

208

209

Appendix III Hand augering data This appendix contains the hand augering data of the cores that have been conducted within the framework of this study. The data from Smokkelhoek (Appendix I, 21), Spalding (Appendix I, 22) and Stede Broec (Appendix I, 23) can be found in the fact sheets concerning these sites. Key [depth]# maximum depth B brick OM organic matter BO bone P pebble(s) CC charcoal PO pottery (sherds) CL clay layers SAL sand layers Fe iron staining SH shell LC lumps of clay SL silt layers LP lumps of peat 1 Beugen core top of

layer magnetic susceptibility x 10-8 m3/kg

description interpretation

A0 0 dark brown sand + B topsoil 35 mottled light brown grey sand + B

+ CC archaeological layer (plough damage)

65 mottled light brown grey clayey sand

undisturbed

80 light brown grey sand + LC undisturbed 90 clayless sand undisturbed 120# A50 0 30.52 dark brown sand + B + P topsoil 50 14.07 mottled light brown grey sand + B

+ CC (?) archaeological feature

55 7.79 mottled light brown grey sand +B + CC

archaeological layer (plough damage)

70 4.36 light brown grey silty sand + LC undisturbed 105 3.02 light brown grey clayless sand +

LC undisturbed

115 grey clayless sand + LC undisturbed, reducing 120# A100 0 25.86 brown grey sand +B + CC topsoil 40 24.58 brown grey sand +B + CC topsoil /archaeological layer 45 6.07 mottled light brown grey sand archaeological layer (plough damage) 55 4.91 light brown grey sand + LC undisturbed 90 2.9 light brown grey clayless sand undisturbed 120# A150 0 24.01 brown grey clayless sand + B topsoil 35 26.62 brown grey clayless sand +B +P topsoil / archaeological layer 45 11.35 mottled light brown grey sand + B (?) archaeological layer 50 4.41 mottled light brown grey sand (?) undisturbed 75 1.06 light brown grey clayless sand undisturbed 120# A200 0 39.31 brown grey clayless sand topsoil 40 33.98 mottled brown sand archaeological layer (plough damage) 55 8.12 mottled clayless brown sand undisturbed 80 light brown grey clayless sand undisturbed 100 1 light yellow grey clayless sand undisturbed 120#

210

Beugen continued A250 0 38.15 brown grey clayless sand topsoil 25 30.88 brown sand topsoil 40 19.88 mottled brown clayey sand archaeological layer (plough damage) 60 mottled light brown grey clayless

sand undisturbed

85 1.28 mottled light yellow grey clayless sand

undisturbed

120# A300 0 34.9 brown grey clayless sand + B topsoil 40 30.24 brown grey clayless sand + B topsoil 45 14.12 mottled light brown grey sand + P archaeological layer (plough damage) 65 2.78 mottled light brown grey clayless

sand undisturbed

115 0.77 mottled light brown grey clayless sand

undisturbed

120# X1 0 28.54 brown grey sand + B topsoil 35 34.11 brown grey sand + B topsoil 45 31.07 mottled brown clayey sand +B +

PO + CC archaeological layer (plough damage)

50 35.7 mottled brown clayey sand + B + PO + CC


55 56.28 mottled brown clayey sand +B + PO + CC


60 36.17 mottled brown clayey sand +B + PO + CC


70 194.65 grey sand + CC + PO archaeological layer 75 29.95 grey sand + CC + PO archaeological layer 80 24.33 grey sand + CC + PO archaeological layer 90 13.52 grey sand + CC + PO archaeological layer 100 15.57 mottled clayless grey sand + P (?) undisturbed 110 18.74 clayless grey sand + P undisturbed 120# X2 0 brown grey sand + B topsoil 40 mottled brown clayless sand + B +

CC + P archaeological layer / feature (plough damage)

70 brown clayless sand undisturbed 80 mottled light yellow grey sand undisturbed

100 mottled light yellow grey sand + LC undisturbed

120# 2 Broekpolder core top of

layer magnetic susceptibility x 10-8 m3/kg

description interpretation

A0 0 14.65 light brown grey sand topsoil 10 6.11 brown sandy clay + B + Fe ? archaeological layer 30 grey sandy clay + Fe 40 3.99 grey clayey sand undisturbed 50 5.79 grey clayey sand + CL + Fe undisturbed 90# A18 0 light brown grey sandy clay topsoil 5 21.90 brown peaty sandy clay + CC ? archaeological feature 25 brown peaty sandy clay 50 5.06 grey clayey sand + CL undisturbed 90# A60 0 20.54 brown sandy clay + CC topsoil 20 brown grey sandy clay + Fe ? archaeological layer 30 9.80 grey sandy clay + Fe undisturbed 50 4.89 light brown grey clayey sand + Fe 60#

211

Broekpolder continued A80 0 16.96 brown sandy clay topsoil 25 light brown grey sandy clay + Fe ? archaeological layer 35 4.20 light brown grey clayey sand + Fe undisturbed 55 2.98 grey sand + Fe and SH undisturbed 90# A100 0 21.68 brown sandy clay topsoil 10 5.30 light brown grey sandy clay + Fe ? archaeological layer 50 light brown grey clayey sand + Fe undisturbed 60 4.67 grey sand +Fe and SH undisturbed 95# A120 0 17.23 brown grey sandy clay + Fe topsoil 35 grey sandy clay +Fe ? archaeological layer 50 4.56 light brown grey clayey sand + Fe undisturbed 60 grey sand undisturbed 90# A140 0 20.87 brown grey sandy clay topsoil 25 5.14 grey sandy clay + Fe ? archaeological layer 50 3.94 light brown grey clayey sand + Fe undisturbed 70# A160 0 16.18 brown grey sandy clay topsoil 15 7.92 grey sandy clay + Fe ditch fill 50 grey sand + Fe bottom of ditch 60 4.90 light brown grey sand +Fe undisturbed 70 grey sand + CL + SH undisturbed 95# A180 0 14.82 brown grey sandy clay topsoil 15 7.93 dark grey sandy clay + Fe + OM archaeological feature / layer 45 7.24 light brown grey clayey sand + Fe undisturbed 70# A200 0 17.60 brown grey sandy clay topsoil 15 7.55 dark grey sandy clay + Fe + OM archaeological feature / layer 35 light brown grey sandy clay + Fe undisturbed 40 8.17 light brown grey clayey sand + Fe + SH undisturbed 60# B0 0 17.65 brown grey clayey sand topsoil 15 6.66 mixed light brown grey clayey sand + Fe ? archaeological layer 50 4.39 light brown grey sand + Fe undisturbed 90 light grey sand + Fe undisturbed 95# B20 0 17.37 brown grey clayey sand topsoil 15 4.82 brown grey clayey sand + Fe ? archaeological layer 50 3.90 light brown grey clayey sand + Fe 60# B40 0 14.86 brown grey clayey sand topsoil 10 6.17 mixed brown grey sand ? archaeological feature 50 4.51 light brown grey sand + Fe undisturbed 60 4.89 light grey sand + Fe undisturbed 65# B60 0 brown grey clayey sand topsoil 5 8.33 dark grey sandy clay modern ditch 25 6.66 light brown grey sandy clay modern ditch with cable/pipe 50# B80 0 brown grey sand topsoil 10 9.64 mixed light brown grey sandy clay ? archaeological feature 25 9.01 grey sandy clay ? archaeological feature 30 7.02 mixed light brown grey sandy clay + LC ? archaeological feature 50 light brown grey sand + Fe Undisturbed 65 3.32 light grey sand + SH Undisturbed 75# B100 0 brown grey sand topsoil 5 12.47 mixed light brown grey sandy clay ? archaeological feature 10 13.91 dark grey sandy clay ? archaeological feature 25 9.13 mixed brown grey sandy clay + Fe and OM ? archaeological feature 65 3.38 light grey sand + LC + SH undisturbed 90#

212

8 Harnaschpolder core top of layer magnetic susceptibility

x 10-8 m3/kg description interpretation

A0 0 13.18 brown grey silt topsoil 15 40.16 brown grey silt disturbed 50 7.95 grey sandy clay +SH undisturbed 80# A20 0 16.88 brown grey clayey silt topsoil 20 9.95 grey sandy clay ditch fill 40 9.03 grey sandy clay ditch fill 50 9.23 light brown grey sandy clay undisturbed 75# A40 0 12.15 brown grey clayey silt topsoil 30 7.3 brown grey clayey silt 40 light brown grey clayey silt undisturbed 60 6.23 light brown grey clayey silt + SAL + Fe undisturbed 80# A60 0 13.6 brown grey clayey silt topsoil 30 10.24 brown grey clayey silt 40 7.67 light brown grey sandy silt + SAL + SH 80 light brown grey sandy silt +SAL + SH + Fe 150# A80 0 brown grey clayey silt topsoil 30 10.24 light brown grey clayey silt ?A-horizon 40 8.68 light brown grey clayey silt + SAL + Fe 80# A100 0 19.3 brown grey clayey silt topsoil 50 5.59 light brown grey sandy silt + SAL 90# A120 0 11.64 brown grey clayey silt topsoil + ?arch. feature 45 7.14 light brown grey sandy clay + Fe 60 5.47 light brown grey silt +SAL + Fe 100# A140 0 11.69 brown grey clayey silt topsoil 25 9.52 light brown grey sandy silt + Fe 45 light brown grey silty clay + Fe 50 7.46 light brown grey clayey silt + Fe 100# A180 0 13.85 grey brown clayey silt topsoil 35 12.52 brown grey silty clay ?A-horizon 45 7.49 light brown grey silty clay + Fe + SAL 60 6.3 light brown grey very silty clay + Fe + SAL 100# A200 0 11.51 grey brown clayey silt topsoil 20 11.36 light brown grey silty clay + Fe 30 8.17 50 3.63 light brown grey silty clay + Fe + SAL 140# A240 0 11.65 brown grey silty clay topsoil 20 9.86 blue grey silty clay 40 light brown grey silty clay + Fe 60 4.61 light brown grey silty clay + Fe + SAL 100# A260 0 10.86 brown grey clayey silt topsoil 40 5.67 ligt brown grey silty clay + Fe 100# A280 0 13.95 brown grey clayey silt topsoil 25 10.4 brown grey clayey silt + BO arch. feature 45 light brown grey silty clay + Fe 80 3.17 dark brown peat 88 0.58 light grey sand 100 dark brown peat 120 brown grey clayey peat + SH 130 brown peat 140 light brown grey sand 150#

213

Harnaschpolder continued A300 0 10.74 brown grey silty clay topsoil 20 8.28 grey sandy clay +Fe 40 5.44 brown sandy clay +Fe 50 3.28 brown clayey peat 75 0.48 dark brown peat 100 brown peat 150# B0 0 13.5 brown grey clayey silt topsoil 25 14.14 brown grey clayey silt 40 7.96 light brown very silty clay +Fe 95# B20 0 10.45 brown grey clayey silt topsoil 25 8.87 brown grey very clayey silt 30 6.48 light brown grey sandy clay +Fe 90 3.06 grey clayey silt +Fe +SH 102 brown clayey peat +SH 150# B40 0 13.98 brown grey clayey silt topsoil 15 9.5 grey silty clay 25 light brown grey silty clay 40 7.65 brown peaty clay +Fe 45 4.59 brown clayey peat +SAL +CL 110 4.04 brown peat 150# C0 0 dark brown grey clayey silt topsoil 30 grey clayey silt +Fe 45 light grey silty clay +SAL 70 blue grey silty clay +SAL 105 brown grey silty clay +SAL 120 dark brown grey clayey silt 330# C20 0 dark brown grey clayey silt topsoil 25 grey clayey silt +Fe 40 grey silty clay +Fe 60 grey sandy clay 135 grey brown clayey silt +SAL 295 grey sand 308 brown peat 340# C40 0 disturbed 15 dark brown grey clayey silt topsoil 30 grey silty clay +Fe 60 grey sandy clay +Fe +SAL 115 black grey silty clay +iron sulphides 130 grey brown silty clay +SAL 240 dark brown silty sand 250 dark grey silty clay +SAL 280 brown grey sand +CL 310 brown peat 320# C60 0 dark brown grey clayey silt topsoil 40 grey clayey silt +SH 50 light brown grey silty clay +SH 105 light brown grey clayey silt +Fe +SAL 140 orange grey silty clay 150 blue grey silty clay 155 brown grey silty clay +SAL 220 brown grey silty clay +SAL +LP 308 brown peat 340#

214

Harnaschpolder continued C80 0 dark grey brown clayey silt topsoil 40 light brown grey clayey silt 50 grey clayey silt +Fe +SAL 155 orange grey clayey silt +SAL 165 grey brown clayey silt+SAL 200 grey brown silty sand 270 brown peat 305 grey brown silty clay 310 black grey silty clay +iron sulphides 320 light blue grey silty clay 330# C100 0 dark brown grey clayey silt topsoil 30 light brown grey clayey silt 60 light brown grey clayey silt +Fe 70 light brown grey clayey silt +Fe +SAL 246 dark grey brown clayey silt +SAL 248 grey brown silt +SAL +KL +LP 260 grey silty sand 305 grey silty sand +SH +LP 318 brown peat 335# C120 0 dark grey brown clayey silt topsoil 30 light brown grey clayey silt 40 light brown grey clayey silt +Fe +SAL +SH 80 light brown grey silty sand +Fe 120 light grey brown silty sand +Fe 160 grey brown silty sand +SAL +CL 270 grey brown silty sand +SAL +CL +LP 280 brown peat 320# C140 0 dark brown grey silty clay topsoil 25 light brown grey silty clay 40 grey silty clay +Fe 105 grey brown silty clay +Fe +SAL 155 dark grey silty sand +CL +LP 305 brown peat 340# C160 0 dark grey brown clayey silt topsoil 30 light brown grey silty clay 40 light brown grey clayey silt +Fe +SAL 110 dark grey sandy silt +Fe 120 orange grey clayey silt 125 light brown grey clayey sand +CL +LP 160 light brown grey clayey sand +CL +LP +SH 170 grey sand 330# C190 0 dark grey brown clayey silt topsoil 25 light brown grey clayey silt +Fe 100 light brown grey sand +Fe +CL 145 grey brown sandy clay +Fe +CL 155 grey brown sandy clay +CL +LP 215 grey clayey sand +SH +LP 246 brown peat 295 brown grey silty clay 300# C250 0 dark grey brown clayey silt topsoil 30 grey very clayey silt +Fe 130 grey clayey silt +Fe +SAL 160 brown grey clayey silt +Fe +SAL +CL 165 brown grey clayey sand +SH +LP 200 brown peat 270 brown grey silty clay 275 grey silty clay 290#

215

Harnaschpolder continued C290 0 dark grey brown silty clay topsoil 15 dark grey brown silty clay +Fe 40 light brown grey silty clay +Fe 55 dark brown silty clay +Fe 60 dark brown peat 75 brown peat +Fe +SAL 85 brown peat 155 brown grey silty clay 165 light blue grey silty clay 190 brown peat 200 light blue grey clay 300 light grey sandy clay +SH 340# D0 0 dark grey brown clayey silt topsoil 25 light brown grey clayey silt +Fe 60 grey clayey silt +Fe +SAL 110 orange grey clayey silt +Fe +SAL 150 grey brown clayey silt +SAL 215 grey brown clayey silt +SAL +LP 315# D20 0 dark grey brown clayey silt topsoil 20 dark brown grey clayey silt 30 dark grey clayey silt +Fe 50 light brown grey clayey silt +Fe 90 light brown grey clayey silt +Fe +SAL 140 orange grey clayey silt +Fe +SL 160 grey brown clayey silt +SAL +LP 215 grey brown clayey silt 250 grey sand +SL +SH +LP 305 brown peat 320# D40 0 dark grey brown clayey silt topsoil 25 light brown grey clayey silt +Fe 50 light brown grey clayey silt +Fe +SAL 145 dark grey clayey silt 150 grey brown clayey silt +SAL 300 brown peat 320# D60 0 dark grey brown clayey silt topsoil 30 light brown grey clayey silt +Fe +SAL 145 grey brown clayey silt +Fe +SAL 180 grey brown clayey silt +SAL 270 grey sand 280 brown peat 300 light brown grey silty clay 305 black grey silty clay +iron sulphides 315 grey sandy silt 320 grey sand 340# 10 Heeten core top of layer magnetic susceptibility

x 10-8 m3/kg description interpretation

A0 0 19.64 dark brown sand +OM plaggensoil 47 18.52 brown sand with yellow motteling disturbed 60# A20 0 18.94 dark brown sand +OM plaggensoil 40 25.15 brown sand +Fe B-horizon 60# A40 0 32.55 dark brown sand +OM plaggensoil 38 51.67 brown sand B-horizon 55 108.51 light brown sand BC-horizon 60#

216

Heeten continued A60 0 20.21 dark brown grey sand +OM plaggensoil 30 31.61 brown sand B-horizon 50 54.74 light brown sand BC-horizon 60# A80 0 22.45 dark brown grey sand +OM plaggensoil 28 31.66 brown sand B-horizon 35 65.79 light brown sand BC-horizon 60 19.57 yellow brown sand C-horizon 70# A100 0 18.88 dark brown grey sand +OM plaggensoil 28 25.01 brown sand B-horizon 34 39.98 light brown sand BC-horizon 60 10.86 yellow brown sand C-horizon 65# A120 0 31.50 dark brown grey sand +OM plaggensoil 40 15.10 brown sand disturbed 50 5.19 yellow brown sand C-horizon 60# B80 0 19.59 dark brown grey sand +OM plaggensoil 30 16.86 brown sand ?archaeological feature / B-horizon 40 9.07 yellow brown sand C-horizon 50# D80 0 24.24 dark brown grey sand +OM plaggensoil 55 17.82 brown sand B-horizon / ?ditch 65# X1 0 20.01 dark brown grey sand +OM plaggensoil 25 20.64 brown sand B-horizon 50 8.71 yellow brown sand C-horizon 60# Y1 0 dark brown grey sand +OM plaggensoil 25 brown sand disturbed 30 yellow brown sand C-horizon 40# Z1 0 dark brown grey sand +OM plaggensoil 25 brown sand BC-horizon 30 yellow brown sand C-horizon 40# Z2 0 29.85 dark brown grey sand +OM plaggensoil 25 14.28 brown sand disturbed / ?archaeological feature 35 9.34 yellow brown sand C-horizon 40# 18 Polre core top layer description interpretation 1 0 brown clayey sand +B +CC topsoil 20 dark brown grey clayey sand +BO 35 dark brown sand B-horizon 55 light brown sand C-horizon 70# 2 0 dark brown grey clayey sand +B topsoil 20 dark brown grey clay with slate 35 dark brown grey clay disturbed 45 grey sand +LC disturbed 60 light grey sand 120 light grey clay 130 light brown sand C-horizon 3 0 brown clayey sand +CC topsoil 30 dark brown sand B-horizon 50 light brown sand C-horizon 70#

217

Polre continued 4 0 brown clayey sand +B topsoil 20 light brown grey clay +B and slate 40 light grey clay 50 light brown sand +B 95 grey sand +CC archaeological deposit 110 dark brown sand 180# 5 0 brown clayey sand +B topsoil 20 brown sandy clay +B and slate 35 light brown grey sandy clay 40 grey white sand C-horizon 70# 6 0 grey brown sandy clay +B and slate topsoil 20 grey sandy clay 30 light brown sand +B +CC +LC 50 grey sand +B +PO and slate archaeological deposit 70 dark grey sand +BB +CC archaeological deposit 100 dark grey sandy clay +BB +CC archaeological deposit 140 light grey sandy clay 155 light grey sandy clay with iron sulphides 180 dark grey sandy clay with iron sulphides 200 light grey clay +LP 220 brown peat 280# 7 0 dark brown clayey sand +B +CC topsoil 20 dark brown clay +B 60 grey sand +B +CC archaeological deposit 180 grey sand 210 light brown sand C-horizon 220# 8 0 dark brown clayey sand +B +PO topsoil 25 brick wall / foundation 40# 9 0 dark brown clayey sand +B topsoil 30 brown clayey sand +B +CC and slag archaeological feature, ?furnace 55 brown sand B-horizon 65 light brown sand C-horizon 70#

218

219

References Abdulfattah R, 2006: Possibilities and limitations for the use of geophysical techniques in solving archaeo-

logical problems in The Netherlands. Unpublished MSc thesis, Vrije Universiteit, Amsterdam. Aitken MJ, 1961: The proton magnetometer survey at Arras (Yorks), Antiquaries Journal 41, 58-9. Anderson K, 1996: De toepsbaarheid van twee geofysische exploratiemethoden voor archeologische doeleinden

binnen Nederland, unpublished PhD thesis, Universiteit van Amsterdam, Amsterdam. Atkinson RJC, 1953: Field archaeology, Methuen, London. De Bakker CE & H Rogaar 1993: Roodzand op de Veluwe, Grondboor & Hamer 47, 24-8. Bartington G & CE Chapman 2004: A high-stability fluxgate magnetic gradiometer for shallow geophysical

survey applications, Archaeological Prospection 11, 19-34. Bates MR & CR Bates 2000: Multidisciplinary approaches to the geoarchaeological evaluation of deeply strati-

fied sedimentary sequences: Examples from Pleistocene and Holocene deposits in southern England, United Kingdom, Journal of Archaeological Sciences 27, 845-58.

Batt C, S Fear & C Heron 1995: The role of magnetic susceptibility as a geophysical survey technique: A site assessment at High Cayton, North Yorkshire, Archaeological Prospection 2, 179-96.

Becker H, JWE Fassbinder, W Irlinger & K Leidorf 1996: Archäologische Prospektion und Luftbildarchäo-logie, Karl Lipp Verlag, München.

Berendsen HJA, 1997: Landschappelijk Nederland, Van Gorcum, Assen. Berner RA, 1984: Sedimentary pyrite formation: An update, Geochimica et Cosmochimica Acta 48, 605-15. Black GA & RB Johnston 1962: A test of magnetometry as an aid to archaeology, American Antiquity 28, 199-

205. Le Borgne E, 1955: Susceptibilité magnétique anormale du sol superficiel, Annales de Géophysique 11, 399-

419. Le Borgne E, 1965: Les propriétés magnétiques du sol. Application a la prospection des sites archéologiques,

Archaeo-Physica. Technische und naturwissenschaftliche Beiträge zur Feldarchäologie, Böhlau Verlag, Köln, Graz.

Brongers JA, 1976: Air photography and Celtic field research in The Netherlands, Nederlandse Oudheden 6, Amersfoort.

Canfield DE, R Raiswell & SH Bottrell 1992: The reactivity of sedimentary iron minerals toward sulfide, American Journal of Science 292, 659-68.

Clark A, 1996: Seeing beneath the soil, Batsford, London. Conyers LB & D Goodman 1997: Ground Penetrating Radar: An introduction for archaeologists, AltaMira

Press, Walnut Creek, London and New Delhi. Cornell RM & U Schwertmann 2003: The iron oxides, Wiley - VCH Verlag GmbH& Co. KGaA, Weinheim. Crawford OGS, 1928: Air survey and archaeology, Southampton. Crockford RH & IR Willett 1995: Drying and oxidation effects on the magnetic properties of sulfidic material

during oxidation, Australian Journal of Soil Research 33, 19-29. Crowther J & P Barker 1995: Magnetic Susceptibility: Distinguishing Anthropogenic Effects from the Natural,

Archaeological Prospection 2, 207-15. Dalan RA & SK Banerjee 1998: Solving archaeological problems using techniques of soil magnetism, Geo-

archaeology 13, 3-36. David A, 1995: Geophysical Survey in Archaeological Field Evaluation, English Heritage Research and Pro-

fessional Services Guideline No. 1, English Heritage, London. David A, 2001: Overview - The role and practice of archaeological prospection. In: DR Brothwell & AM

Pollard (eds.), Handbook of archaeological sciences, John Wiley & Sons, Ltd., Chichester. David A & A Payne 1997: Geophysical surveys within the Stonehenge landscape: A review of past endeavour

and future potential. In: B Cunliffe & C Renfrew (eds.), Science and Stonehenge, Oxford University Press, Oxford, 73-114.

Dearing JA, JA Hannam, AS Anderson & EMH Wellington 2001: Magnetic, geochemical and DNA properties of highly magnetic soils in England, Geophysical Journal International 144, 183-96.

220

Dearing JA, KL Hay, SMJ Baban, AS Huddleston, EMH Wellington & PJ Loveland 1996: Magnetic suscepti-bility of soil: an evaluation of conflicting theories using a national data set, Geophysical Journal Inter-national 127, 728-34.

Decker KV & I Scollar 1962: Iron Age square enclosures in Rhineland, Antiquity 36, 175-8. Deeben J, B Groenewoudt, D Hallewas & W Willems 1999: Proposals for a practical system of significance

evaluation in archaeological heritage management, European Journal of Archaeology 2, 177-99. Deeben J, B Groenewoudt, D Hallewas, C van Rooijen & P Zoetbrood 2005: Op zoek naar de archeologische

voorraad. In: MH van den Dries & WJH Willems (eds.), Innovatie in de Nederlandse Archeologie. Liber amicorum voor Roel W. Brandt, SIKB, Gouda, 37-54.

Dekkers MJ, HF Passier & MAA Schoonen 2000: Magnetic properties of hydrothermally synthesized greigite (Fe3S4) - II. High- and low-temperature characteristics, Geophysical Journal International 141, 809-19.

Dent DL & LJ Pons 1995: A world perspective on acid sulphate soils, Geoderma 67, 263-76. Dockrill SJ & JA Gater 1992: Tofts Ness: Exploration and interpretation in a prehistoric landscape. In: P

Spoerry (ed.), Geoprospecting the archaeological landscape, Oxford. Dockrill SJ & IA Simpson 1994: The identification of prehistoric anthropogenic soils in the Northern Isles using

an integrated sampling methodology, Archaeological Prospection 1, 75-92. Doner HE & WC Lynn 1989: Carbonate, halide, sulfate, and sulfide minerals. In: Dixon, JB & SB Weed (eds.),

Minerals in the soil environment, Soil Science Society of America, Madison. Doneus M, W Neubauer & A Eder-Hinterleitner 1997: 3D-combination of ground magnetics and aerial photo-

graphs in archaeological prospection, Annales Geophysicae, Part I Solid Earth Geophysics & Natural Hazards 15 (Supplement I), C83.

Ellis C & AG Brown 1998: Archaeomagnetic dating and palaeochannels sediments: Data from the Mediaeval channel fills at Hemington, Leicestershire, Journal of Archaeological Sciences 25, 149-63.

Fassbinder JWE & WE Irlinger 1994: Aerial and magnetic prospection of an 11-13th century motte in Bavaria, Archaeological Prospection 1, 65-9.

Fassbinder JWE & H Stanjek 1993: Occurence of bacterial magnetite in soils from archaeological sites, Archaeologia Polona 31, 117-28.

Fassbinder JWE, H Stanjek & H Vali 1990: Occurence of magnetic bacteria in soil, Nature 343, 161-3. Flamman JP & Besselsen (in prep.): Archeologische Monumentenzorg AHR-Project Rapportage Archeologische

Monumentenzorg.. Flannery KV 1976: Sampling by intensive surface collection. In: KV Flannery (ed.), The early Mesoamerican

village, Academic Press, New York, 51-62. Flannery KV 1982: The golden Marshalltown, American Anthropologist 84, 265-78. Frijtag Drabbe CAJ von 1947: De luchtfoto in dienst van de archeologie, Een kwart eeuw oudheidkundig

bodemonderzoek in Nederland. gedenkboek AE van Giffen, Meppel, 491-6. Gaffney VL, CF Gaffney & M Corney 1998: Changing the Roman landscape: the role of geophysics and remote

sensing. In: J Bayley (ed.), Science in archaeology, an agenda for the future, English Heritage, London. Gaffney C & J Gater 2003: Revealing the buried past; geophysics for archaeologists, Tempus, Stroud. Geel B van & GJ Borger 2002: Sporen van grootschalige zoutwinning in de kop van Noord-Holland, Wester-

heem 51, 244-60. Gibbs-Eggar Z, B Jude, J Dominik, J-L Loizeau & F Oldfield 1999: Possible evidence for dissimilatory bacterial

magnetite dominating the magnetic properties of recent lake sediments, Earth and Planetary Science Letters 168, 1-6.

Groenewoudt BJ, 1994: Prospectie, waardering en selectie van archeologische vindplaatsen: een beleidsge-richte verkenning van middelen en mogelijkheden, Nederlandse Archeologische Rapporten 17, Amersfoort.

Hansel CM, SG Benner & S Fendorf 2005: Competing Fe(II)-induced mineralization pathways of ferrihydrite, Enviromental Science & Technology 39, 7147-53.

Haselgrove C, 1985: Inference from ploughsoil artefact samples. In: C Haselgrove, M Millett & I Smith (eds.), Archaeology from the ploughsoil: studies in the collection and analysis of field survey data, University Press, Sheffield, 7-29.

Heeringen RM van, A Smit & EM Theunissen 2004: Archaeology in the future: baseline measurement of the physical quality of the archaeological monument at Broekpolder, Rijksdienst voor het Oudheidkundig Bodemonderzoek, Amersfoort.

Heron CP & CF Gaffney 1987: Archaeogeophysics and the site: Ohm sweet Ohm? In: CF Gaffney & VL Gaffney (eds.), Pragmatic archaeology, theory in crisis? BAR British Series, Oxford.

Heslop D, MJ Dekkers, PP Kruiver & IHM van Oorschot 2002: Analysis of isothermal remanent magnetisation acquisition curves using the expectation-maximisation algorithm, Geophysical Journal International 148, 58-64.

Hesse A 1962: Geophysical prospecting for archaeology in France, Archaeometry 5, 123-5.

221

Hey G & M Lacey 2001: Evaluation of Archaeological Decision-making Processes and Sampling Strategies, Planarch project 1, Oxford Archaeological Unit / English Heritage and Kent County Council.

Hissel M & H van Londen 2004: de kwaliteit van de waarneming; een vergelijking van boormethoden voor archeologisch inventariserend veldonderzoek, AAC/projectenbureau, Amsterdam.

Hunt CP, BM Moskowitz & SK Banerjee 1995: Magnetic properties of rocks and minerals. In: TJ Ahrens (ed.), Rock Physics and Phase Relations, AGU, Washington DC, 189-204.

Ikeya M, T Oka & T Omura 1997: Evaluation of Environment using Electron Spin Resonance (ESR): Micros-cope Images of Gypsum (CaSO4•2H2O) Microcrystals in Borehole Cores at Konya Basin, Turkey, Japan Review 8, 193-208.

IMAGO-projectgroep 2003: IMAGO Eindrapportage. Lelystad, RDIJ rapporten 2003-133, 2003-13a en 2003-13b.

Jansen B, 2003: Plangebied Smokkelhoek, gemeente Kapelle. Een inventariserend archeologisch onderzoek, RAAP notitie 494, RAAP, Amsterdam.

Jenness D, 1930: Archaeological field work in North America during 1929, American Anthropologist, new series 32, 342-74.

Kattenberg AE, 1999: The detection of buried mudbrick structures by the means of geophysical methods: A survey in Euesperides, Libya. University of Bradford, Department of Archaeological Sciences.

Kattenberg AE, 2007: The mapping of a peat-extraction landscape, windows into a past landscape by combining geophysical methods and coring. In: AMJ de Kraker & GJ Borger (eds.), Veen-Vis-Zout, landschappelijke dynamiek in de zuidwestelijke delta van de Lage Landen (Geoarchaeological and Bioarchaeological Studies 8), 35-44.

Kattenberg AE & G Aalbersberg 2004: Archaeological prospection of the Dutch perimarine landscape by means of magnetic methods, Archaeological Prospection 11, 227-35.

Kintigh KW, 1988: The effectiveness of subsurface testing: a simulation approach, American Antiquity 53, 686-707.

Kley J van der 1968: De ontdekking van het Romeinse castellum Nigropullo bij Zwammerdam, Westerheem 17, 117-25.

Kley J van der 1970: Opnieuw een Romeins castellum gelocaliseerd door systematisch bodemonderzoek, Westerheem 19, 22-33.

Kluiving SJ, N Brand & GJ Borger 2006:De West-Brabantse delta: Een verdronken landschap vormgeven. Geoarchaeological and Bioarchaeological Studies 7, Amsterdam.

Kostka JE & GW Luther 1995: Seasonal cycling of Fe in salt-marsh sediments, Biochemistry 29, 159-81. Krakker JJ, MJ Shott & PD Welch 1983: Design and evaluation of shovel test sampling in regional archaeo-

logical survey, Journal of Field Archaeology 10, 469-80. Kruiver PP, MJ Dekkers & D Heslop 2001: Quantification of magnetic coercivity components by the analysis of

acquisition curves of isothermal remanent magnetisation, Earth and Planetary Science Letters 189, 269-76. Kruiver PP, CG Langereis, MJ Dekkers & W Krijgsman 2003: Rock -magnetic properties of multi-component

natural remanent magnetisation in alluvial red beds (NE Spain), Geophysical Journal International 153, 317-32.

Kruiver PP & HF Passier 2001: Coercivity analysis of magnetic phases in sapropel S1 related to variations in redox conditions, including an investigation of the S-ratio, Geochemistry, Geophysics, and Geosystems Paper number 2001GC000181.

Langeveld MCM, RM Jayasena & JP Flamman 2003: Begraven Beugen. Inventariserend veldonderzoek op het toekomstig bedrijvenpark Beugen-Zuid, gemeenteBoxmeer, AAC publicaties, Amsterdam.

Leckebusch J, 2003: Ground-penetrating radar: A modern three-dimensional prospection method, Archaeologi-cal prospection 10, 213-40.

Lerici CM, 1961: Methods used in the archaeological prospecting of Etruscan tombs, Studies in Conservation 6, 1-8.

Linford N, 1994: Mineral magnetic profiling of archaeological sediments, Archaeological Prospection 1, 37-52. Linford NT, 2004: Magnetic ghosts: mineral magnetic measurements on Roman and Anglo-Saxon graves,

Archaeological Prospection 11, 167-80. Linford NT, 2006: The application of geophysical methods to archaeological prospection, Reports on progress

in physics 69, 2205-57. Linford NT & MG Canti 2001: Geophysical evidence for fires in antiquity: preliminary results from an

experimental study; paper given at the EGS XXIV General Assembly in The Hague, April 1999, Archaeo-logical Prospection 8, 211-25.

Lovley DR, SJ Giovannoni, DC White, JE Champine, EJP Phillips, A Gorby & S Goodwin 1993: Geobacter metallireducens gen. nov. sp. nov., a microorganism capable of coupling the complete oxidation of organic compounds to the reduction of iron and other metals, Archives of microbiology 159, 336-44.

222

Lovley DR, JF Stolzt, JGL Nord & EJP Phillips 1987: Anaerobic production of magnetite by a dissimilatory iron-reducing microorganism, Nature 330, 252-4.

Luther GW, A Giblin, RW Howarth & RA Ryans 1982: Pyrite and oxidized iron mineral phases formed from pyrite oxidation in salt marsh and estuarine sediments, Geochimica and Cosmochimica Acta 46, 2665-9.

Luther GW & JE Kostka 1992: Seasonal iron cycling in the salt-marsh sedimentary environment - the important of ligand complexes with Fe(II) and Fe(III) in the dissolution of Fe(III) minerals and pyrite, respectively, Marine Chemistry 40, 81-103.

Maher BA, 1998: Magnetic properties of modern soils and Quaternary loessic paleosols: paleoclimatic impli-cations, Palaeogeography, Palaeoclimatology, Palaeoecology 137, 25-54.

Maher BA & RM Taylor 1988: The formation of ultrafinegrained magnetite in soils, Nature 336, 368-70. Maki D, JA Homburg & SD Brosowske 2006: Thermally activated mineralogical transformations in archaeolo-

gical hearths: inversion from maghemite gamma-Fe2O4 phase to haematite alpha-Fe2O4 form, Archaeolo-gical Prospection 13, 207-27.

Marmet E, M Bina, N Fedoroff & A Tabbagh 1999: Relationships between human activity and the magnetic properties of soils: a case study in the medieval site of Roissy-en-France, Archaeological Prospection 6, 161-70.

McOmish D, D Field & G Brown 2002: The field archaeology of the Salisbury plain training area, English Heritage, Swindon.

Mulder EFJ de, MC Geluk, I Ritsema, WE Westerhoff & TE Wong 2003: De ondergrond van Nederland, NITG-TNO, Utrecht.

Mullender TAT, AJ van Velzen & MJ Dekkers 1993: Continuous drift correction and separate identification of ferromagnetic and paramagnetic contributions in thermomagnetic runs, Geophysical Journal International 114, 663-72.

Mullins CE, 1977: Magnetic susceptibility of the soil and its significance in soil science - a review, Journal of Soil Science 28, 223-46.

Oonk S, 2006: Chemische prospectie Tiel-Passewaaij. In: Heeren, S (ed.), Opgraving bij Tiel-Passewaaij 1, De nederzetting aan de Passewaaijse Hogeweg, Zuidnederlandse Archeologische Rapporten, ACVU, Amster-dam.

Orbons J, 1998: Is meten weten? Geofysisch onderzoek op het terrein van de afgebrande LBK boerderij in Archeon, Bulletin voor archeologische experimenten en educatie 3, 3-8.

Peters C & R Thompson 1998: Magnetic identification of selected natural iron oxides and sulphides, Journal of Magnetism and Magnetic Materials 183, 365-74.

Poulton SW, MD Krom & R Raiswell 2004: A revised scheme for the reactivity of iron (oxyhydr)oxide minerals towards dissolved sulfide, Geochimica et cosmochimica acta 68, 3703-15.

Rabenhorst MC, 1990: Micromorphology of induced iron sulfide formation in a Chesapeake Bay (USA) tidal marsh. In: LA Douglas (ed.), Soil micromorphology: a basic and applied science, Elsevier, New York.

Raiswell R & DE Canfield 1998: Sources of iron for pyrite formation in marine sediments, American Journal of Science 298, 219-45.

Reynolds JM, 1997: An introduction to applied and environmental geophysics, John Wiley & Sons, Chichester. Ritsema CJ & JE Groenenberg 1993: Pyrite oxidation, carbonate weathering, and gypsum formation in a drained

Potential Acid Sulfate Soil, Soil Science Society of America Journal 57, 968-76. Scollar I, A Tabbagh & A Hesse 1990: Archaeological prospection and remote sensing, Cambridge University

Press, Cambridge. Sier MM, 2003: Ellewoutsdijk in de Romeinse tijd. ADC-Rapporten, Bunschoten. Sueur C, 2006: Remote sensing voor archeologische prospectie en monitoring. RAAP-rapport 1261, RAAP

Archeologisch Adviesbureau & Vestigia Archeologie en Cultuurhistorie, Amsterdam/Amersfoort. Taylor RM, 1980: Formation and properties of Fe(II) and Fe(III) hydroxy carbonate and its possible significance

in soil formation, Clay minerals 15, 369-82. Terkorn L, 1991: de empirische cyclus in werking: het Assendelver polder-project. In: JHF Bloemers & T van

Dorp (eds.), Pre- en protohistorie van de lage landen, Open Universiteit, Houten. Terkorn L, M Diepeveen-Jansen, S Gerritsen, J Kaarsemaker, M Kok, L Kubiak-Martens, J Slopsma & P Vos

(in press): Landscapes in the Broekpolder: excavations around a monument with aspects of the Bronze Age tot the Modern.

Thompson R & F Oldfield 1986: Environmental magnetism, Allen & Unwin, London. Tol A, P Verhagen, A Borsboom & M Verbruggen 2004: Prospectief boren, een studie naar de betrouwbaarheid

en toepasbaarheid van booronderzoek in de prospectiearcheologie. RAAP-rapport 1000, RAAP Archeolo-gisch Adviesbureau, Amsterdam.

Topografische Dienst 1990: Foto-atlas Noord-Holland, Robas, Den Ilp.

223

Velzen AJ van & JDA Zijderveld 1992: A method to study alterations of magnetic minerals during thermal demagnetization applied to a fine-grained marine marl (Trubi Formation, Sicily), Geophysical Journal International 110, 79-90.

Vliet AP van 2007: De Zeeuwse drie-eenheid bedreigd. Vis, veen en zout in de 16e en 17e eeuw. In: AMJ de Kraker & GJ Borger (eds.), Veen-Vis-Zout, landschappelijke dynamiek in de zuidwestelijke delta van de Lage Landen (Geoarchaeological and Bioarchaeological Studies 8), 55-64.

Vos P & P Kiden, J Deeben, E Drenth, M-F van Oorsouw & L Verhart (eds.) 2005: De landschapsvorming tijdens de steentijd, Meppel.

De Vries-Metz WH, 1993: Mogelijkheden en resultaten van archeologische Remote Sensing in een verdwijnend prehistorisch cultuurlandschap, Proefschrift Universiteit van Amsterdam, Amsterdam.

Waldus WB & HM van der Velde 2006: Archeologie in vogelvlucht. Toepassingsmogelijkheden van het AHN in de archeologie (Geoarchaeological and Bioarchaeological Studies 6), Amsterdam.

Weston D, 1996: Soil science and the interpretation of archaeological sites: A soil survey and magnetic suscepti-bility analysis of Altofts `Henge', Normanton, West Yorkshire, Archaeological Prospection 3, 39-50.

Weston DG, 1999: The physical and chemical property analysis of archaeological soils and sediments;a study of the influence of pedological processess upon the magnetic properties of soil iron and upon soil phos-phorus and the implications for prospection methods and site analysis. Unpublished PhD thesis, Depart-ment of Archaeological Sciences. University of Bradford, Bradford.

Weston DG, 2001: Alluvium and geophysical prospection, Archaeological Prospection 8, 265-72. Weston DG, 2002: Soil and susceptibility: aspects of thermally induced magnetism within the dynamic pedo-

logical system, Archaeological Prospection 9, 207-15. Weston DG, 2004: The influence of waterlogging and variations in pedology and ignition upon resultant sus-

ceptibilities: a series of laboratory reconstructions, Archaeological Prospection 11, 107-20. Weston DG, 2004: Laboratory tests on magnetic susceptibility influences, Archaeological Prospection 11, 107-

20. Wijngaarden M van, LB Venema, RJ de Meijer, JJG Zwolsman, B van Os & JMJ Gieske 2002: Radiometric

sand – mud characterisation in the Rhine – Meuse estuary. Part A. Fingerprinting, Geomorphology 43, 87-101.

Willems WJH, 1997: Archaeological heritage management in the Netherlands: Past, present and future. In: WJH Willems, H Kars & DP Hallewas (eds.), Archaeological Heritage Management in the Netherlands: fifty years state service for archaeological investigations, Van Gorcum, Assen.

Willems WJH, 2008: Archaeological resource management and preservation, in: H Kars & RM van Heeringen (eds.) Preseving archaeological remains in situ. Proceedings of the 3rd conference 2006, Amsterdam (Geoarchaeological and Bioarchaeological Studies 10), 253-60, Amsterdam.

Zagwijn WH, 1991: Nederland in het Holoceen, Rijks Geologische Dienst, Haarlem.

224

225

Samenvatting Hoofdstuk 1 Introductie Deze studie beoogt te onderzoeken of magnetische methoden ingezet kunnen worden in het kader van de Archeologische Monumentenzorg (AMZ) in Nederland. Magnetische methoden zijn geofysische, non-destructieve prospectieve methoden, welke buiten Nederland veelvuldig worden ingezet in het vakgebied van de archeologische prospectie, zowel in de wetenschappelijke als in de commerciële archeologie. In Nederland echter, hebben geofysische onderzoeksmethoden geen plaats verworven binnen de archeologie. In dit proefschrift wordt onderzocht: - of magnetische methoden in Nederland gebruikt kunnen worden om archeologische sporen en vind-

plaatsen in kaart te brengen; - of magnetische prospectie in Nederland op landschapsschaal kan worden ingezet; - in welke delen van Nederland magnetische methoden wel of niet inzetbaar zijn; - of aan de hand van bodemmonsters voorspeld kan worden of met behulp van magnetische methoden

een bepaalde archeologische vindplaats in kaart gebracht kan worden. Magnetische methoden kunnen gebruikt worden om archeologische structuren als greppels en kuilen in kaart te brengen, maar niet alle structuren blijken magnetisch detecteerbaar te zijn. Of er al dan niet magnetisch contrast is ontstaan, is grotendeels afhankelijk van het type bodemmateriaal (lithologie, organisch materiaal, bodemvormende en post-depositionele processen) waarin archeologische sporen en lagen zijn ingebed. Voor deze studie zijn 29 archeologische vindplaatsen in drie verschillende afzettingsmilieus onderzocht: in getijdenafzettingen, eolische afzettingen en rivierafzettingen. Parallel hieraan zijn een aantal voor de Nederlandse prospectieve archeologie relevante thema's onderzocht, onder andere de magnetische prospectie van verdronken dorpen, off-site structuren en archeologische vindplaatsen onder esdekken. Hoofdstuk 2 Archeologische prospectie in Nederland In dit hoofdstuk wordt archeologische prospectie geïntroduceerd aan de hand van het concept van archeologische voorraad. Met behulp van non-destructieve onderzoekstechnieken kan kennis gegene-reerd worden over het bodemarchief. Op deze wijze wordt een deel van de onbekende archeologische voorraad deel van de bekende archeologische voorraad. Deze kennis is nodig om beslissingen te kun-nen nemen binnen de cyclus van de AMZ, bijvoorbeeld over de waardering en selectie van archeo-logische vindplaatsen. In het tweede deel van het hoofdstuk wordt de geschiedenis van de prospectieve archeologie in Neder-land beschreven. Aanvankelijk richtte men zich met luchtfotografie en veldverkenningen vooral op het onderzoek van oppervlakkige archeologische verschijnselen. Na de introductie van de grondboor in de Nederlandse archeologie in de jaren tachtig van de twintigste eeuw kon ook prospectief onder-zoek onder de oppervlakte uitgevoerd worden. In dezelfde periode werden geofysische methoden in de Nederlandse archeologie geïntroduceerd, deze zijn echter nooit een geïntegreerd onderdeel van de prospectieve methodologie geworden. In de overige Europese landen en de Verenigde Staten ontwik-kelde de prospectieve archeologie zich op een andere wijze. Hier werd juist wel veelvuldig gebruik gemaakt van geofysische technieken en werd de geofysica geïntegreerd in het universitaire onderzoek.

226

In de jaren negentig veranderde de focus van prospectief onderzoek in het algemeen, en van de geofysica in het bijzonder, van de vindplaats naar het archeologische landschap. Door de technische ontwikkelingen werd de geofysische apparatuur zodanig verbeterd dat er grote oppervlaktes in kaart gebracht konden worden. De afwijkende ontwikkeling van de prospectieve archeologie in Nederland valt mogelijk te verklaren uit een combinatie van de focus van het prospectief onderzoek op de grondboormethode en een gebrek aan inbedding van archeologische geofysica in universiteiten en diensten. Als gevolg hiervan hebben geofysische methoden geen duidelijke plaats gekregen in de KNA, en worden ze vrijwel niet ingezet binnen de cyclus van de AMZ. Hoofdstuk 3 De toepasbaarheid van magnetische methoden in Nederland; de onderliggende principes In dit hoofdstuk wordt uitgelegd waarom archeologische sporen en structuren doorgaans magnetisch verschillen van de matrix waarin ze zijn ingebed. Vervolgens wordt beschreven hoe deze magnetische variaties een afwijking (anomalie) in het aardmagnetisch veld kunnen veroorzaken, die met behulp van een magnetometer gemeten kan worden. De magnetische susceptibiliteit is een maat voor het gemak waarmee een object gemagnetiseerd kan worden. De magnetische susceptibiliteit van de bodem wordt voornamelijk bepaald door de hoeveel-heid en het type ijzeroxide dat erin voorkomt. Archeologische lagen en de bouwvoor hebben over het algemeen een verhoogde magnetische susceptibiliteit ten opzichte van de ongestoorde bodemlagen. Deze verhoging kan bijvoorbeeld veroorzaakt worden door verhitting van het bodemmateriaal in het verleden, of door de werking van bacteriën. Als archeologische sporen opgevuld raken met materiaal met een verhoogde magnetische susceptibiliteit kunnen ze, onder invloed van het aardmagnetisch veld, een afwijking (anomalie) in het aardmagnetisch veld veroorzaken (geïnduceerd magnetisme). Remanent magnetisme bestaat ook zonder de aanwezigheid van een extern veld, zoals het aard-magnetisch veld, en is onafhankelijk van geïnduceerd magnetisme. Terwijl laatstgenoemde vooral voorkomt in de vulling van archeologische sporen zoals kuilen en greppels, hebben objecten en archeologische structuren die tot hoge temperatuur verhit zijn geweest, als haarden, ovens, maar ook bakstenen, een remanent magnetisme. Of er sprake is van remanent of geïnduceerd magnetisme kan vaak afgeleid worden van de vorm, grootte en oriëntatie van een magnetische anomalie. Post-deposi-tionele processen zoals langdurige waterverzadiging van de bodem en gley kunnen de ijzersamen-stelling van de bodem veranderen, en daarmee de magnetische susceptibiliteit. Hoofdstuk 4 Methodologie Tijdens dit onderzoek zijn binnen Nederland 27 en buiten Nederland twee archeologische vind-plaatsen onderzocht. Op sommige vindplaatsen is een magnetometer survey uitgevoerd voorafgaand aan een archeologische opgraving. Deze procedure kon echter niet steeds worden toegepast, op andere locaties werd het magnetisch onderzoek juist voorafgegaan door een proefsleuf- of booronderzoek. Bemonstering van het bodemmateriaal werd, waar mogelijk, verricht tijdens de opgraving. Op plaatsen waar geen opgraving (meer) werd uitgevoerd gebeurde dit met behulp van een grondboor. Het geheel van onderzochte vindplaatsen is geografische gespreid over Zuid- en Midden-Nederland (zie Figuur 12), en beslaat de archeologische perioden van de Bronstijd tot en met de post-Middel-eeuwse periode. Het grootste deel van de onderzochte terreinen kan archeologisch gezien als neder-zetting gekarakteriseerd worden. Daarnaast is aandacht besteed aan off-site structuren, begraaf-plaatsen, industriële vindplaatsen en infrastructurele objecten. Tijdens het onderzoek is gebruik gemaakt van de Geoscan FM36 fluxgate gradiometer magnetometer. Onderzoek naar de magnetische susceptibiliteit van het bodemmateriaal is uitgevoerd met een Agico KLY-2 magnetische susceptibiliteitsbrug en met een Bartington MS2B susceptibiliteitsmeter. In dit hoofdstuk is beschreven hoe het verhittingsexperiment is uitgevoerd, alsook de thermomagnetische, ARM en IRM metingen. De software die gebruikt is voor het verwerken van de magnetische data in Geoplot en Archeosurveyor.

227

Hoofdstuk 5 Getijdenafzettingen Dit hoofdstuk, bestaande uit drie delen, beschrijft de mogelijkheden en de moeilijkheden van de magnetische prospectie van archeologische vindplaatsen op getijdenafzettingen. In het eerste deel worden de resultaten getoond van het onderzoek naar de detecteerbaarheid van archeologische structuren die met veenwinning (moernering) te maken hebben. Onderzoek in Smokkelhoek en Kolhorn heeft aangetoond dat magnetometrie en weerstandsonderzoek goede methodes zijn om elementen in een veenwinningslandschap gedetailleerd in kaart te brengen. Hierdoor kan prospectief onderzoek, bijvoorbeeld booronderzoek, naar pre-moerneringsstructuren gerichter worden uitgevoerd. Het tweede deel van het hoofdstuk wordt beschreven dat uit dit onderzoek blijkt dat de vulling van archeologische sporen in getijdenafzettingen geen of nauwelijks magnetisch contrast heeft met de omliggende matrix. Twee vindplaatsen worden besproken: Harnaschpolder en Smokkelhoek. Behalve het gebrek aan contrast, is ook de magnetische susceptibiliteit van de bodemmonsters bijzonder laag. In scherp contrast komen op beide vindplaatsen grote, sterk magnetische anomalieën voor die waarschijnlijk veroorzaakt zijn door de aanwezigheid van de tijdens het booronderzoek aangetoonde (ferrimagnetische) ijzersulfiden die onder de grondwaterspiegel in de ondergrond aanwezig zijn. Deze ijzersulfiden komen voornamelijk voor in getijdenafzettingen en worden gevorm uit een combinatie van sulfaat uit zeewater, ijzer in oplossing en organisch materiaal. In deze sectie van het hoofdstuk wordt gesuggereerd dat zowel het voorkomen van contrastloze sedimenten als het ontstaan van zeer magnetische structuren met een geologische oorsprong wellicht veroorzaakt kunnen zijn door de inundatie van de matrix met zeewater. In het derde deel van dit hoofdstuk wordt verslag gedaan van het gedetailleerde magnetische onderzoek dat is uitgevoerd op monsters van Harnaschpolder en Broekpolder. Hierbij is onderzocht wat de oorzaak is van de afwezigheid van een magnetisch contrast op vindplaatsen in getijden-afzettingen. Tevens diende het onderzoek om vast te stellen of de grote magnetische anomalieën inderdaad veroorzaakt worden door de aanwezigheid van ijzersulfiden in de ondergrond. Behalve magnetische susceptibiliteitsmetingen zijn thermomagnetische en Isothermal Remanent Magnetisation metingen uitgevoerd. Het resultaat van dit deel van het onderzoek is dat het waarschijnlijk is dat de magnetische susceptibiliteit van de sedimenten tijdens de inundatie met zeewater gehomogeniseerd is, daarnaast lijken ook gley en de uitspoeling van ijzer uit de bodem een homogeniserend effect te hebben gehad op de ijzermineralogie van de sedimenten. Uit de laag die de sterke magnetische anomalieën veroorzaakt, zijn monsters genomen die zowel greigiet als pyriet bevatten. Hieruit blijkt dat de anomalieën inderdaad veroorzaakt worden door de onder de grondwaterspiegel aanwezige ijzersulfiden. Hoofdstuk 6 Eolische en rivierafzettingen In de dekzandgebieden van Oost- en Zuid-Nederland zijn zeer lage magnetische susceptibiliteits-waardes gemeten. Op de vindplaatsen die bemonsterd zijn, bijvoorbeeld Breda en Den Dolder, is er een goed magnetisch contrast tussen de bouwvoor en de onderliggende lagen, maar een gebrek aan contrast tussen de archeologische sporen en de matrix waarin ze zijn ingebed. Deze resultaten van het laboratoriumonderzoek komen overeen met het gebrek aan archeologische anomalieën in de magnetometer data op de vindplaatsen van Breda en Slabroek. Hoewel magnetometrisch onderzoek in het algemeen minder goed geschikt lijkt voor prospectief archeologisch onderzoek op grove, minerale zanden kan de methode in specifieke situaties wel ingezet worden. Op de vindplaats Heeten Hordelman bijvoorbeeld, konden een aantal ijzerovens en afvalkuilen in kaart gebracht worden. Naast het gebrek aan magnetisch contrast hebben ook de aanwezigheid van een plaggendek en het voorkomen van roodzand een negatieve invloed op de bruikbaarheid van magnetometrie in dekzand-gebieden. De vindplaats Limmen, in duinzand, liet juist een zeer goed magnetisch contrast zien voor een gedeelte van de archeologische sporen, voornamelijk greppels en waterputten, waarvan een aanzien-lijk aantal tijdens de magnetometer survey in kaart gebracht konden worden. In het lössgebied werd eveneens slechts één vindplaats onderzocht, een Romeinse villa in Meerssen.

228

Op deze vindplaats kon een gedeelte van de sporen magnetisch in kaart gebracht worden, en een magnetisch contrast tussen de sporen en de matrix werd in het laboratorium bevestigd. In tegenstelling tot de dekzanden lijken archeologische vindplaatsen in fijner bodemmateriaal (duinzand en löss) wel goed geschikt voor magnetometrisch onderzoek. Ook in het rivierengebied kon tijdens dit onderzoek op een gedeelte van de vindplaatsen een gedeelte van de archeologische sporen in kaart gebracht worden met behulp van een magnetometeronderzoek. In het rivierengebied in Midden-Nederland ontbrak een magnetisch contrast op de vindplaatsen Deil en Zaltbommel, terwijl in Wijk bij Duurstede juist grote magnetische contrasten gemeten zijn. Waar deze verschillen door veroorzaakt worden is onduidelijk. Twee vindplaatsen in de Maasvallei, Beugen en Borgharen, lieten grote magnetische contrasten zien, die goed interpreteerbare magnetische anoma-lieën veroorzaakten. Anderson had in een niet gepubliceerd onderzoek in Gennep, eveneens in de Maasvallei, al eerder een correlatie tussen archeologische sporen en de resultaten van een magneto-meteronderzoek vastgesteld. Hoofdstuk 7 Magnetometer respons van archeologische structuren in Nederland Een verzameling voorbeelden van de magnetometrische respons van veel voorkomende archeo-logische sporen bepalen de inhoud van dit hoofdstuk. De magnetische data zijn als uitsnede van de originele dataset gepresenteerd, en per groep geordend. De behandelde categorieën betreffen neder-zettingen (kuilen, greppels, muren en waterputten), off-site (ploegsporen, greppels, drenkplaatsen), grafstructuren (graven, kringgreppels), industrieel (moerneringsputten, legakkers en ovens) en infra-structuur (wegen). Dit hoofdstuk kan bij toekomstig magnetometrisch onderzoek in Nederland gebruikt worden als catalogus bij de interpretatie van nieuwe data. Hoofdstuk 8 Discussie Hoofdstuk 8 begint met de kritische beschouwing van de methodologie van het onderzoek waar dit proefschrift verslag van doet. De onderzoeksstrategie is tijdens het onderzoek aangepast, wat uiteinde-lijk tot een ander resultaat geleid heeft dan oorspronkelijk verwacht was. De grootste verandering is de veelheid van kleine gebieden die onderzocht zijn in tegenstelling tot drie grotere gebieden. Daarnaast is het resultaat van het onderzoek in zekere mate afhankelijk van de keuze van het instrumentarium en van de bemonsteringsstrategie, de gevolgen hiervan worden in dit hoofdstuk besproken. De contrasten in de magnetische susceptibiliteit tussen archeologische lagen en vullingen van sporen met de ongestoorde matrix zijn afhankelijk van een aantal factoren. Ten eerste de aanwezigheid van organisch materiaal van belang voor de potentiële verhoging van magnetische susceptibiliteit. De bouwvoor en de vulling van archeologische sporen bevatten vaak meer organisch materiaal dan de ongestoorde matrix: dit gaat bij de onderzochte vindplaatsen veelal samen met een verhoogde susceptibiliteit. Voor bacteriën dient organisch materiaal als voedingsstof, en als elektron acceptor. Daarnaast zorgt organisch materiaal tijdens verbranding voor een reducerend milieu, hetgeen de ver-hoging van de magnetische susceptibiliteit ten goede komt. Ook de ijzersulfiden die tijdens deze studie in de getijdenafzettingen aangetroffen zijn kwamen voor in combinatie met organisch mate-riaal, vermoedelijk hebben deze ijzersulfiden een bacteriële oorsprong. Ten tweede is de korrelgrootte is van invloed op de magnetische susceptibiliteit van het bodemmateriaal. In deze studie bleek dat in grofzandige deposities weinig magnetisch contrast aanwezig was, terwijl het bodemmateriaal met een silt – klei samenstelling juist veel contrast te zien gaf, met uitzondering van die gebieden waar een inundatie van zeewater plaatsgevonden heeft. De zuurgraad van de bodem kan ook van invloed zijn op de magnetische susceptibiliteit. IJzeroxiden in zeer zure gronden kunnen verplaatst of uitgespoeld worden, hetgeen invloed heeft op de magnetische susceptibiliteit. De laatste factor die hier genoemd wordt is het post-depositionele proces, bijvoorbeeld gley, waterverzadiging en de inundatie van zeewater. Al deze processen veroorzaken een verandering in de magnetische susceptibiliteit van de bodem.

229

De getijdenafzettingen worden vooral gedomineerd door de post-depositionele processen, gley, de uitspoeling van ijzer, en de inundatie met zeewater. In de dekzandgebieden lijken voornamelijk de grofkorrelige structuur van het bodemmateriaal en de podzolisatie van de bodem onder invloed van een lage pH een negatieve invloed te hebben gehad op het behoud van magnetische contrasten. In het duinzand lijkt dit niet het geval, al zijn hier zijn wel magnetische contrasten gemeten. De mogelijkheid bestaat dat dit contrast afkomstig is van onder de grondwaterspiegel. In het lössgebied speelt de korrelgrootte in combinatie met het organisch materiaal waarschijnlijk een belangrijke rol in de conservering van het magnetisch contrast, net als in het rivierengebied. Thermoremanente objecten en archeologische sporen zoals bakstenen, ovens en haarden veroorzaken altijd een magnetische anomalie in het aardmagnetisch veld en zijn in principe altijd meetbaar. Hoofdstuk 9 Conclusie Hoofdstuk 9 begint met een samenvatting van de onderliggende principes van magnetische prospectie in de archeologie. Daarna wordt de bruikbaarheid van de methode voor de verschillende onderzochte geogenetische gebieden besproken. Het gebruik van de magnetometer als prospectiemethode op getijdenafzettingen wordt op basis van deze studie afgeraden. Een uitzondering hierop vormen archeologische sporen die zich onder de grondwaterspiegel bevinden en organisch materiaal bevatten omdat deze door de formatie van ijzersulfiden juist een groot magnetisch contrast kunnen krijgen. Ook het dekzandgebied lijkt niet geschikt voor de magnetische prospectie van archeologische resten. In het duin- en lössgebied is de magnetometer waarschijnlijk wel goed bruikbaar. Het rivierengebied, voornamelijk in de Maasvallei en op een aantal vindplaatsen daarbuiten, bevat relatief veel archeo-logische sporen die in kaart gebracht zijn. De Maasvallei zou een logisch beginpunt zijn voor verder magnetometeronderzoek in Nederland. De invloed van de maskering van het onderzoeksobject door overliggende lagen, zoals een plaggen-dek, resulteert in een grotere afstand tussen de magnetometer en de archeologische resten. Deze afstandsvergroting veroorzaakt een reductie van het signaal en heeft een negatieve invloed op de detecteerbaarheid van archeologische sporen. De variabiliteit van de niet-archeologische lagen in de matrix kan ervoor zorgen dat een aanwezig archeologisch signaal niet in de data te onderscheiden is. Voorbeelden van lagen met een variabele magnetische susceptibiliteit zijn de bouwvoor, het plaggen-dek in afdekkende lagen en het voorkomen van ijzersulfiden en roodzand in onderliggende lagen. De hypothese wordt geponeerd dat de formatie van ijzersulfiden in de getijdenafzettingen samengaat met een verandering van de magnetische susceptibiliteit van het bodemmateriaal in de geïnundeerde gebieden. Wanneer de grondwaterspiegel weer gedaald is en de archeologische resten zich hierboven bevinden, is het magnetisch contrast dat hiervoor mogelijk bestaan heeft uitgewist. Er is meer onder-zoek nodig om deze hypothese te bevestigen. Gebaseerd op de tijdens dit onderzoek opgedane kennis, wordt een plan voorgesteld voor de gefa-seerde introductie van magnetometrie in de Nederlandse Archeologische Monumentenzorg. Hierbij moeten geofysische methoden beter geïntegreerd worden in de KNA, waarbij de Maasvallei als startpunt kan dienen voor de systematische toepassing van magnetometrie in het prospectieonderzoek. Magnetometeronderzoek kan geen vervanging zijn van de andere prospectiemethoden, maar wel een aanvulling. Het niet benutten van een bruikbare techniek werkt het verlies van informatie in de hand. De sterke punten van magnetometrie zijn de goede resolutie en de snelheid van de methoden (tussen booronderzoek en proefsleufonderzoek in) en de resulterende horizontale dataset, waarin individuele grondsporen te onderscheiden zijn. De combinatie van magnetometrie en grondbooronderzoek ligt dan ook voor de hand: met behulp van de magnetometer kunnen grondsporen in kaart gebracht worden, terwijl booronderzoek gebruikt kan worden om deze sporen te dateren, de diepte te bepalen en om uitspraken te kunnen doen over de conserveringstoestand. In sommige gevallen kan magneto-metrie in plaats van of in combinatie met proefsleuven of opgravingen uitgevoerd worden, vooral daar waar alleen informatie over de locatie van bepaalde sporen nodig is. Ook verstoringen van het bodemarchief kunnen met behulp van de magnetometer in kaart gebracht worden. Met andere woorden, de magnetometer kan in veel situaties ingezet worden, maar niet in alle. Met deze conclusie heeft dit proefschrift een bijdrage geleverd aan de kaders waarbinnen deze nieuwe techniek succesvol geïntegreerd kan worden in de archeologische monumentenzorg.

230

Documents

UvA-DARE (Digital Academic Repository) The Application of ... · 1.3 A D utch perspective 3 1.4 T his study 4 2 A rch aeological p rosp ection in T h e N eth erlan d s 7 2.1 Introduction